US7849122B2 - Self-authenticating quantum random bit generators - Google Patents

Abstract

Various embodiments of the present invention are directed to self-authenticating, quantum random bit generators that can be integrated into an optoelectronic circuit. In one embodiment, a quantum random bit generator comprises a transmission layer that includes an electromagnetic radiation source coupled to a waveguide branching into a first, second, and third waveguides. The radiation source generates pulses of electromagnetic radiation in a first polarization state. Polarization rotators are operably coupled to the second and third waveguides and rotate pulses transmitted in the second waveguide into a second polarization state and rotate pulses transmitted in the third waveguide into a third polarization state. The system control generates a sequence of bits based on polarization basis states of the pulses transmitted in the first waveguide, and tomographically authenticates randomness of the sequence based on polarization basis states of the second and third pulses.

Description

CROSS-REFERENCES TO RELATED APPLICATIONS

This application is a continuation-in-part to application Ser. No. 11/407,513, filed Apr. 20, 2006 and application Ser. No. 11/546,158 filed Oct. 10, 2006.

TECHNICAL FIELD

The present invention relates to random number generators, and, in particular, to methods and systems for generating a sequence of random bits using quantum mechanical properties of quantum systems.

BACKGROUND OF THE INVENTION

Random numbers have applications in numerous areas including game playing, statistical sampling, evaluating integral equations, particle transport calculations, and computations in statistical physics, just to name a few. As a result, random number generators (“RNGs”) figure prominently in methods and systems that use random numbers. For example, RNGs are key components of secure systems and are used extensively to generate keys for cryptography. An ideal RNG generates numbers which cannot be predicted in advance, cannot be reliably reproduced, and are distributed according to a uniform distribution. In other words, RNGs ideally generate a sequence of unbiased random numbers. However, many commonly used RNGs either generate sequences of seemingly random numbers or may be susceptible to generating biased sequences of numbers.

RNGs have been implemented in software to generate sequences of seemingly random numbers using formulas and numerical methods. Software-based RNGs are in general formula-based RNGs and are referred to as “pseudorandom number generators” because the formulas allow for prediction and reproduction of a sequence of pseudorandom numbers, provided the same initial parameters are used. A recursive Lehmer pseudorandom number generator (“LPNG”) is an example of a commonly used pseudorandom number generator is given by:
x _n+1 =Ax _n +C(mod M)
where

x_n is the nth number of a sequence of random numbers; and

A, C, and M are parameters that can be adjusted to ensure that a sequence of numbers generated by the LPNG appears random.

Typically, M is assigned the word size of a computer employed to compute a sequence of pseudorandom numbers, and x₀, the seed, is assigned a prime number. For example, assigning A, C, and M the values 21, 1, and 32 (5 bits), respectively, and assigning x₀ the prime number 13, the LPNG generates the following sequence of pseudorandom integers: 13, 18, 27, 24, 25, 14, 7, etc. Alternative approaches may seed a pseudorandom number generator with the time produced by a computer-system clock each time the pseudorandom number generator is initiated. However, even using the time provided by a system clock is not infallible because one may be able to determine the time when the pseudorandom number generator was initiated.

Hardware-based RNGs have also been developed to generate sequences of random numbers based on chaotic fluctuations observed in thermal noise generated by atomic, molecular, and electrical systems. For example, thermal noise generated by an electric current flowing through a resistor can be used as a source of a sequence of random numbers by assigning numbers to the magnitude of voltage equilibrium fluctuations. However, hardware-based RNGs are not always reliable because the systems employed are susceptible to environmental changes. For example, an electric noise-based RNG can be biased by changing the temperature of the system. In addition, the methods typically employed to authenticate the randomness of a sequence generated by a hardware-based RNG are deterministic software-based methods, which can be used to determine whether the sequence is statistically well-behaved but cannot be used to evaluate the true randomness of the sequence. Further, sufficiently powerful data processing systems with appropriate models or algorithms may become able to predict chaotic or thermal processes, even if only for a short time.

Quantum random bit generators (“QRBGs”) are another type of hardware-based RNG. A bit is the most basic unit of information used in computing and information processing and exists in one of two states represented by binary numbers “0” and “1.” QRBGs are based on quantum-mechanical properties of substantially identical quantum systems. A sequence of random numbers can be generated by associating each number with the outcome of a measurement performed on a quantum system. The numbers generated in this manner are truly random because each measurement projects the state of a quantum system onto one of many possible states at the time the measurement is performed, and, according to the standard interpretation of quantum mechanics, no amount of refinement of the measurement methods and measuring devices can overcome the uncertainty in the outcome of a measurement performed on a quantum system. As a result, QRBGs are highly desirable systems for generating sequences of random numbers.

Quantum measurements can be used to generate random bits. For example, a sequence of random bits can be generated by detecting the transmission and reflection of a 45°-polarized photon on a polarizing beamsplitter with two photomultiplier detectors located in the output channels of a polarizing beamsplitter. Each detector has the same probability of registering a detection event, but one cannot predict which detector records the next detection event. By assigning the binary number “0” to a detection at one of the detectors and the binary number “1” to a detection at the other detector, a sequence of random binary numbers can be constructed. A sequence of bits can be used to generate a sequence of random integers. For example, suppose separately transmitting 30 45°-polarized photons to a polarizing beamsplitter generates the following sequence of random bits:

• • 000110101011100101010111100100
    This sequence can be partitioned into 5-bit words to give a sequence of random base 2 numbers 00011, 01010, 11100, 10101, 01111, and 00100, which can then be translated into a corresponding sequence of random base 10 integers 3, 10, 28, 21, 15, and 4, respectively.

Although QRBGs appear to offer a convenient method and system for generating a sequence of random numbers, the QRBGs may be susceptible to generating sequences of biased bits by tampering with the photon source. In addition, methods typically employed to authenticate the randomness of bit sequences generated by QRBGs are deterministic software-based methods, which are unreliable for assessing whether or not a sequence of bits is truly random. Physicists, cryptographers, computer scientists, and quantum-information users have recognized a need for QRBGs that can be used to reliably generate sequences of random bits, can be integrated into optoelectronic devices, and can also detect, authenticate, and correct biases in the sequences of random bits using methods that rely on the non-deterministic properties of quantum systems.

SUMMARY OF THE INVENTION

Various embodiments of the present invention are directed to self-authenticating, quantum random bit generators that can be integrated into an optoelectronic circuit. In one embodiment of the present invention, a self-authenticating, quantum random bit generator comprises a transmission layer and a system control. The transmission layer includes an electromagnetic radiation source coupled to a waveguide branching into a first waveguide, a second waveguide, and a third waveguide. The electromagnetic radiation source is also configured to generate pulses of electromagnetic radiation in a first polarization state. One or more polarization rotators are positioned and configured to rotate pulses transmitted in the second waveguide into a second polarization state and rotate pulses transmitted in the third waveguide into a third polarization state. The system control is configured to generate a sequence of bits based on polarization basis states of the pulses transmitted in the first waveguide, and tomographically authenticates randomness of the sequence of bits based on polarization basis states of the pulses transmitted in the second and third waveguides.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 illustrates a cubic cavity.

FIG. 2 illustrates a three-dimensional coordinate system with two independent polarization vectors and a normalized wavevector as basis vectors.

FIG. 3 illustrates a representation of electric and magnetic field components of an electromagnetic wave in the coordinate system shown in FIG. 2.

FIG. 4 is an energy-level diagram of quantized electromagnetic radiation.

FIG. 5 illustrates a probability distribution associated with a pulse output from a source and transmitted in a waveguide to a detector.

FIGS. 6A-6B show plots of vertical and horizontal polarization basis states.

FIG. 7 illustrates a Poincare sphere representation of polarization states.

FIGS. 8A-8D show plots of four polarization states.

FIG. 9 illustrates a geometric representation of Stokes parameters.

FIG. 10 illustrates a general schematic representation of a quantum random number generator that represents an embodiment of the present invention.

FIG. 11 illustrates a schematic representation of a first quantum random bit generator that represents an embodiment of the present invention.

FIGS. 12A-12B illustrate a slotted polarization rotator that represents an embodiment of the present invention.

FIGS. 13A-13B illustrate a slanted-ridge polarization rotator that represents an embodiment of the present invention.

FIGS. 14A-14C illustrate schematic representations of three different random bit generators, each representing an embodiment of the present invention.

FIG. 15 illustrates a hypothetical sequence of raw counts that may be recorded by a system control that represents an embodiment of the present invention

FIG. 16 illustrates generating a sequence of random bits from a sequence of raw counts that represents an embodiment of the present invention.

FIG. 17 illustrates the quantum random number generator shown in FIG. 10 for an adversarial scenario.

FIG. 18 is a plot of a min-entropy that represents an embodiment of the present invention.

FIG. 19 shows a control-flow diagram that represents one of many embodiments of the present invention for generating sequences of random bits.

FIG. 20 shows a control-flow diagram for the routine “generate sequence of raw binary numbers {right arrow over (r)}_n” called in step 1906 of FIG. 19 and represents one of many embodiments of the present invention.

FIG. 21 shows a control-flow diagram for the routine “tomographic analysis” called in step 1908 of FIG. 19 and represents one of many embodiments of the present invention.

FIG. 22 is a control-flow diagram for the routine “sift the sequence of raw binary numbers” called in step 1910 of FIG. 19 and represents one of many embodiments of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

Various embodiments of the present invention are directed to self-authenticating quantum random bit generators (“QRBGs”) that can be integrated into optoelectronic devices. System embodiments of the present invention generate sequences of random bits and include quantum-mechanical-based methods that can be used to evaluate and authenticate randomness of the sequence and remove biased bits from the sequence. System embodiments are also small enough to be included in optoelectronic devices, such as personal computers, servers, and portable electronic devices, just to name a few.

Embodiments of the present invention are mathematical in nature and, for this reason, are described below with reference to numerous equations and numerous graphical illustrations. Although mathematical expressions, alone, may be sufficient to fully describe and characterize embodiments of the present invention to those skilled in the art of quantum optics and quantum information, the more graphical, problem oriented examples, and control-flow-diagram approaches included in the following discussion are intended to illustrate various embodiments of the present invention in a variety of different ways so that the present invention may be accessible to readers with various backgrounds. Also, in order to assist the reader in understanding descriptions of various embodiments of the present invention, overview subsections of related topics in physics are provided. In a first subsection, an overview of quantum mechanics is provided. An overview of electromagnetic radiation and quantum optics is provided in a second subsection. An overview of coherent states is provided in a third subsection. An overview of polarization states and Stokes parameters is provided in a fourth subsection. Finally, various system and method embodiments of the present invention are described in a fifth subsection.

Overview of Quantum Mechanics

Embodiments of the present invention employ concepts in quantum mechanics. The textbook “Quantum Mechanics, Vol. I and II,” by Claude Cohen-Tannoudji, Bernard Diu and Frank Laloe, Hermann, Paris, France, 1977, is one of many references for the field of quantum mechanics. In this subsection, topics in quantum mechanics that relate to embodiments of the present invention are described. Additional details can be obtained from the above-referenced textbook, or from many other textbooks, papers, and journal articles related to quantum mechanics.

Quantum mechanics models the observed behavior, atomic and subatomic levels, of systems comprising photons, electrons, atoms, and molecules. Quantum systems exist in discrete states that are characterized by discrete measurable quantities. A state of a quantum system is represented by a ket and is denoted |Ψ

where Ψ is a label that represents a state of a quantum system. For example, an electron has two intrinsic spin-angular-momentum states that correspond to two measurable spin-angular-momentum values

/2 and

/2, where

 is approximately 1.0546×10⁻³⁴ Js. The spin state that corresponds to the spin-angular momentum

/2 is referred to as “spin up” and can be denoted |↑

and the spin state that corresponds to the spin angular momentum

/2 is referred to as “spin down” and can be denoted |↓

Various different labels can be assigned to various different quantum states. For example, the spin up and spin down states |↑

and |↓

can also be represented by the kets |½

and |−½

respectively. Also, a single label can be used to represent different states in entirely different quantum systems. For example, the ket “|1

can represent a first quantized vibrational level of a diatomic molecule and can also be used to represent a single photon, as described below, in a following subsection.

A measurement employed to determine a measurable quantity of a quantum system is represented by an operator {circumflex over (Ψ)}, where the symbol “^” denotes an operator. In general, an operator operates on a ket from the left as follows:
{circumflex over (Ψ)}(|Ψ

)={circumflex over (Ψ)}|Ψ

where {circumflex over (Ψ)}|Ψ

is a ket representing an observed quantum state. Typically, an operator {circumflex over (Ψ)} is associated with a set of states called “eigenstates.” An eigenstate is represented as “|ψ_i

with the following property:
{circumflex over (Ψ)}|ψ_i

=ψ_i|ψ_i

where i is a non-negative integer; and ψ_i is a real value, called an “eigenvalue,” that corresponds to a discrete real-valued measurable quantity that is observed when the quantum system is in the eigenstate |ψ_i

For example, a measurement employed to determine the spin angular momentum of an electron parallel to the z-axis is represented by Ŝ_z, and the eigenvalue-eigenstate representations of observed spin-angular-momentum values are:

${\hat{S}}_z\uparrow \rangle =\frac{\mathrm{<figure-callout\; id="2"\; label="ħ"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">\hslash </figure-callout>}}{2}\uparrow \rangle ,\mathrm{and}$ ${\hat{S}}_z\downarrow \rangle =-\frac{\mathrm{<figure-callout\; id="2"\; label="ħ"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">\hslash </figure-callout>}}{2}\downarrow \rangle$

The eigenstates of an operator are complex vectors that span a complex vector space called a “state space.” The eigenstates constitute a basis of the vector space if every state belonging to the state space has a unique linear superposition on the basis. For example, a state |↑

in a state space spanned by the N eigenstates {|ψ_i

of an operator {circumflex over (Ψ)} the can be written as a linear superposition of the eigenstates as follows:

$\Psi \rangle =\sum _{i=1}^N{c}_i{\Psi }_i\rangle$
where c_i is a complex valued coefficient called the “amplitude.” The state space associated with an operator is also called a “Hilbert space.” A Hilbert space includes a mathematical operation called the “inner product.” The inner product of two states |↑

and |Ξ

is represented by:

Ξ|Ψ

where

Ξ| is called a “bra,” and represents the complex conjugate and transpose of the state |Ξ

The inner product has the following property:

Ξ|Ψ

=

Ψ|Ξ

where “*” represents the complex conjugate. The basis eigenstates of a Hilbert space are orthonormal, or in mathematical notation:

ψ_i|ψ_j

=δ_ij
where δ_ij is 1 when i equals j, and 0 otherwise. For example, the inner product of the eigenstates of a single electron Hilbert space are:

↑|↑

=

↓|↓

=1, and

↑|↓

=

↓|↑

=0

The orthonormality property of the eigenstates of a Hilbert space can be used to determine the coefficients of the linear superposition of states |Ψ

Taking the inner product of |Ψ

with

ψ_j| gives the corresponding coefficient:

$\langle {\psi }_j❘\Psi \rangle =\sum _{i=1}^N{c}_i\langle {\psi }_j❘{\psi }_i\rangle =\sum _{i=1}^N{c}_i{\delta }_{\mathrm{ij}}=c_j$
Substituting for the coefficients in the linear superposition gives:

$\Psi \rangle =\sum _{i=1}^N{\psi }_i\rangle \langle {\psi }_i❘\Psi \rangle$
Because |Ψ

is an arbitrary ket in the Hilbert space,

$\sum _{i=1}^N{\psi }_i\rangle \langle {\psi }_i=\hat{1}$
where “{circumflex over (1)}” is the identity operator. The summation is called the “completeness relation,” and the eigenstates {|ψ_i

are said to be “complete.”

Prior to a measurement corresponding to an operator {circumflex over (Ψ)}, a quantum system can be thought of as simultaneously existing in all of the eigenstates {|ψ_i

of the operator {circumflex over (Ψ)}, which is represented by the (pure state) linear superposition of states:

$\Psi \rangle =\sum _{i=1}^N{c}_i{\psi }_i\rangle =\sum _{i=1}^N{\psi }_i\rangle \langle {\psi }_i|\Psi \rangle$
The measurement corresponding to the operator {circumflex over (Ψ)} projects the quantum system initially in the state |Ψ

onto one of the eigenstates |ψ_i

In other words, a measurement on a quantum system is essentially a filtering process that places the state of the quantum system into one of the eigenstates in the linear superposition at the time of the measurement. For example, an electron with an unknown spin orientation prior to a measurement corresponding to the operator Ŝ_z is represented in a linear superposition of states:
|Ψ

=c₁|↑

+c₂|↓

A spin determination measurement Ŝ_z projects the state of the electron onto either the state |↑

or the state |↓

at the time of the measurement. In other words, just after the spin determination measurement, the electron is either in the state |↑

or the state |↓

There is a corresponding irreversible change in the state of a quantum system as a result of a measurement. Irreversibility can only be avoided when the quantum system is already in one of the quantum states before the measurement is performed. As a result, one cannot infer the prior state of a quantum system based on the outcome of a single measurement. For example, if the outcome of a spin measurement is

/2, it is not possible to determine whether the system was already in the state |↑

or in a linear superposition of the spin states |↑

and |↓

at the time of the measurement.

Although it is not possible to know in advance which of the various states |ψ_i

the state of a quantum system will be projected onto, the probability of the quantum system being found in a particular state |ψ_i

immediately after the measurement is given by:
Pr(ψ_i)=|c _i|²=

ψ_i|Ψ

where |Ψ

is normalized, and |c_i|² equals c_i*c_i and gives the outcome probability. For example, prior to a spin determination measurement in the spin basis {|↑

|↓

, consider an electron coherently prepared with a ½ probability of being found in the spin state |↑

and a ½ probability of being found in the spin state |↓

The state associated with the electron in such as spin state prior to a spin determination measurement can be represented by:

$\Psi \rangle =\frac{1}{\sqrt{2}}\uparrow \rangle +\frac{1}{\sqrt{2}}\downarrow \rangle$

The expectation value of a measurement performed on a quantum system that is represented by the linear superposition of states |Ψ

is mathematically represented by:

{circumflex over (Ψ)}

=

Ψ|{circumflex over (Ψ)}|

and is determined by applying the completeness relation as follows:

$\begin{array}{c}\langle \hat{\Psi }\rangle =\sum _{i=1}^N\sum _{j=1}^N\langle \Psi |{\psi }_i\rangle \langle {\psi }_i\left|\hat{\Psi }\right|{\psi }_j\rangle \langle {\psi }_j|\Psi \rangle \\ =\sum _{i=1}^N{\psi }_i{\langle {\psi }_i|\Psi \rangle }^2\end{array}$
The expectation value represents the weighted eigenvalue average result expected from measurements on the quantum systems in the ensemble, where the initial state |Ψ

of the quantum system is the same for each member of the ensemble. In other words, the linear superposition of states representing each quantum system is identical prior to the measurement. In practice, such an ensemble could be realized by preparing many identical and independent quantum systems all in the same state, or by repeatedly preparing a single system in the same state. Note that the expectation value may not be the value obtained for each measurement and, therefore, is not to be confused with the eigenvalue obtained from the measurement. For example, the expectation value of Ŝ_z can be any real value between the eigenvalues

/2 and

/2, but the actual measured value of Ŝ_z for an electron is always either

/2 or

/2 in each individual measurement.

The expectation value of a single quantum system in a state |Ψ

can also be described using a density operator defined by:
{circumflex over (ρ)}=|Ψ

Ψ|
where the state |Ψ

is also called a “pure state,” which is distinguished from a statistical mixture of states described below. The density operator is represented in the {|ψ_i

basis by a matrix called the “density matrix” whose matrix elements are:
ρ_ij=

ψ_i|{circumflex over (ρ)}|ψ_j

=c _i *c _j
The density operator characterizes the state of the quantum system. In other words, the density operator provides all the physical information that can be calculated from the state |Ψ

For example, the sum of the diagonal matrix elements of the density matrix is given by:

$\sum _i{c_i}^2=\sum _i{\rho }_{\mathrm{ii}}=\mathrm{Tr}\left(\rho \right)=1$
where Tr represents the trace, or sum of the diagonal elements, of a matrix. For example, the density matrix of a two-state quantum system in the pure state:
|Ψ

=c ₁|ψ₁

+c ₂|ψ₂

is given by:

$\rho =\left[\begin{array}{cc}{\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">c</figure-callout>}}_1{c}_1^{*}& {\mathrm{<figure-callout\; id="1"\; label="c"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">c</figure-callout>}}_1{c}_2^{*}\\ {\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">c</figure-callout>}}_2{c}_1^{*}& {\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">c</figure-callout>}}_2{c}_2^{*}\end{array}\right]$
where the diagonal elements are the probabilities associated with projecting the quantum system into either the state |ψ₁

or the state |ψ₂

and the off-diagonal elements represent the interference effects between the states |ψ₁

and |ψ₂

In addition, the expectation value of a quantum system in the state |Ψ

can be expressed as:

$\begin{array}{c}\langle \hat{\Psi }\rangle =\sum _{i,j}\langle {\psi }_j|\Psi \rangle \langle \Psi |{\psi }_i\rangle \langle {\psi }_i\left|\hat{\Psi }\right|{\psi }_j\rangle \\ =\sum _{i,j}\langle {\psi }_j\left|\hat{\rho }\right|{\psi }_i\rangle \langle {\psi }_i\left|\hat{\Psi }\right|{\psi }_j\rangle \\ =\mathrm{Tr}\left\{\hat{\rho }\hat{\Psi }\right\}\end{array}$

However, it is often the case that information about a quantum system is incomplete. For example, a quantum system can be in any one of the states |Ψ₁

|Ψ₂

|Ψ₃

. . . , each with an associated probability p₁, p₂, p₃, . . . , where the probabilities satisfy the conditions:

$0\le {\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">p</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">p</figure-callout>}}_2,p_3,\dots \le 1,\mathrm{and}\sum _i{p}_i=1$
The quantum system is said to exist in a “statistical mixture of states.” The density operator for a statistical mixture of states can be determined as follows. As described above, the probability that a measurement of the observable {circumflex over (Ψ)} on a quantum system in the pure state |Ψ_i

yields a result ψ_n is:
Pr _i(ψ_n)=

Ψ_i|ψ_n

ψ_n|Ψ_i

=|

ψ_n|Ψ_i

|²
However, the probability Pr_i(ψ_n) of observing ψ_n in a statistical mixture of states is weighted by p_i and the summed over i to give:

$\begin{array}{c}\mathrm{Pr}\left({\psi }_n\right)=\sum _i{p}_i{\mathrm{Pr}}_i\left({\psi }_n\right)\\ =\sum _i{p}_i\langle {\psi }_n|{\Psi }_i\rangle \langle {\Psi }_i|{\psi }_n\rangle \\ =\langle {\psi }_n\left|\hat{\rho }\right|{\psi }_n\rangle \mathrm{where}\end{array}$ $\hat{\rho }=\sum _i{p}_i{\Psi }_i\rangle \langle {\Psi }_i$
is the density operator associated with a statistical mixture of states. The associated density matrix elements are given by:

$\begin{array}{c}{\rho }_{\mathrm{np}}=\langle {\psi }_n\sum _i{p}_i{\Psi }_i\rangle \langle {\Psi }_i||{\psi }_p\rangle \\ =\sum _i{p}_i{c}_n^{\left(i\right)}{c}_p^{{\left(i\right)}^{*}}\end{array}$
The physical meaning of the density matrix is described for a two-state quantum system comprising a mixture of states:
|Ψ_i

=c ₁ ⁽ⁱ⁾|ψ₁

+c ₂ ⁽ⁱ⁾|ψ₂

The corresponding density matrix is given by:

$\rho =\left[\begin{array}{cc}{\mathrm{<figure-callout\; id="11"\; label="\rho "\; filenames="US07849122-20101207-D00006.png"\; state="\{\{state\}\}">\rho </figure-callout>}}_{11}& {\mathrm{<figure-callout\; id="12"\; label="\rho "\; filenames="US07849122-20101207-D00002.png"\; state="\{\{state\}\}">\rho </figure-callout>}}_{12}\\ {\rho }_{21}& {\rho }_{22}\end{array}\right]=\left[\begin{array}{cc}\sum _i{p}_i{c}_1^{\left(i\right)}{c}_1^{{\left(i\right)}^{*}}& \sum _i{p}_i{c}_1^{\left(i\right)}{c}_2^{{\left(i\right)}^{*}}\\ \sum _i{p}_i{c}_2^{\left(i\right)}{c}_1^{{\left(i\right)}^{*}}& \sum _i{p}_i{c}_2^{\left(i\right)}{c}_2^{{\left(i\right)}^{*}}\end{array}\right]$
The diagonal matrix elements can be interpreted to mean that when the state of the quantum system is |Ψ_i

the diagonal matrix element ρ₁₁ represents the average probability of finding the quantum system in the state |ψ₁

and the diagonal matrix element ρ₂₂ represents the average probability of finding the quantum system in the state |ψ₂

When the same measurement is carried out N times under identical conditions, Nρ₁₁ will be found in the state |ψ₁

and Nρ₂₂ will be found in the state |ψ₂

The off-diagonal elements ρ₁₂ and ρ₂₁ express the average interference effects between the states |ψ₁

and |ψ₂

Note that unlike the diagonal matrix elements, the off-diagonal matrix elements can be zero even though neither of the products c₁ ⁽ⁱ⁾c₂ ⁽ⁱ⁾* and c₂ ⁽ⁱ⁾c₁ ⁽ⁱ⁾* is zero, which means that the average over N measurements has cancelled out the interference effects of the states |ψ₁

and |ψ₂

A tensor product is a way of combining Hilbert spaces of different quantum systems to form Hilbert spaces that represent combined quantum systems. For example, H_Ψ is a Hilbert space of a first quantum system, and H_Ξ is a Hilbert space of a second quantum system. The Hilbert space denoted by H_Ψ

 H_Ξ represents a combined Hilbert space, where the symbol

 represents a tensor product. The operators {circumflex over (Ψ)} and {circumflex over (Ξ)} correspond to the Hilbert spaces H_Ψ and H_Ξ, respectively, and each operates only on the corresponding eigenstates as follows:
({circumflex over (Ψ)}

{circumflex over (Ξ)})(|ψ

|ξ

)=({circumflex over (Ψ)}|ψ

)

({circumflex over (Ξ)}|ξ

)
where |ψ

represents a state in the Hilbert space H_Ψ, and |ξ

represents a state in the Hilbert space H_Ξ. The tensor product |ψ

⊕|ξ

can be abbreviated as |ψ

|ξ

, |ψ,ξ

or |ψξ

For example, the spin states of two electrons in an atomic orbital are bases for a combined Hilbert space. The two electrons can either both be spin up, both be spin down, the first electron spin up and the second electron spin down, or the first electron spin down and the second electron spin up. The various tensor product representations of two spin up electrons are given by:
|↑

₁

|↑

₂=|↑

₁|↑

₂=|↑,↑

₁₂=|↑↑

₁₂
where the subscripts 1 and 2 refer to the first and second electrons.

An Overview of Electromagnetic Radiation and Quantum Optics

In this subsection, a brief description of electromagnetic radiation and quantum optics that relates to embodiments of the present invention is described. Quantum optics is a field of physics that relates the application of quantum mechanics to electromagnetic radiation. Electromagnetic radiation confined to a cavity with perfectly reflecting walls is quantized. Quantized electromagnetic radiation can be applied to more general unconfined optical systems, such as electromagnetic radiation propagating in free space or in an optical fiber.

Electromagnetic radiation confined to a cavity, with no free charges and currents, comprises an electric field component {right arrow over (E)}({right arrow over (r)},t) and a magnetic field component {right arrow over (B)}({right arrow over (r)},t) that are related in terms of a vector potential {right arrow over (A)}({right arrow over (r)},t) satisfying the wave equation:

${\nabla }^2\stackrel{\rightharpoonup }{A}-\frac{1}{{\mathrm{<figure-callout\; id="2"\; label="c"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">c</figure-callout>}}^2}\frac{{\partial }^2\stackrel{\rightharpoonup }{A}}{\partial {\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">t</figure-callout>}}^2}=0$
and the Coulomb, non-relativistic gauge condition:
∇·{right arrow over (A)}({right arrow over (r)},t)=0
where the electric and magnetic field components are determined by:

$\stackrel{\rightharpoonup }{E}\left(\stackrel{\rightharpoonup }{r},t\right)=-\frac{\partial \stackrel{\rightharpoonup }{A}\left(\stackrel{\rightharpoonup }{r},t\right)}{\partial t},\mathrm{and}$ $\stackrel{\rightharpoonup }{B}\left(\stackrel{\rightharpoonup }{r},t\right)=\nabla \times \stackrel{\rightharpoonup }{A}\left(\stackrel{\rightharpoonup }{r},t\right)$
The electromagnetic radiation is assumed to be propagating and is subject to periodic boundary conditions imposed by a cubic, or quantization, cavity with perfectly reflecting walls, where the walls are of length L. FIG. 1 illustrates a cubic cavity 100. Orthogonal axes 102, 104, and 106 represent the x, y, and z Cartesian coordinate axes. The finite dimensional cubic cavity 100 imposes periodic boundary conditions on solutions to the wave equation. For example, in the x, y, and z-directions, plane wave solutions to the vector potential wave equation satisfy the condition:
exp(i{right arrow over (k)}·{right arrow over (r)})=exp(i{right arrow over (k)}·({right arrow over (r)}+{right arrow over (L)}))
where {right arrow over (L)} is vector (L, L, L), and

{right arrow over (k)} is called the “wavevector” with components:

$\stackrel{\rightharpoonup }{k}=\frac{2\pi }{L}\left(m_x,m_y,m_z\right),$

and

m_x, m_y, and m_z are integers.

Each set of integers (m_x, m_y, m_z) specifies a normal mode of the electromagnetic radiation, and the magnitude of the wavevector {right arrow over (k)}, is called the wavenumber, k, and is equal to ω_k/c, where c represents the speed of light in free space and ω_k is the angular frequency. Note that in real life the spectrum of normal modes of an electromagnetic field is actually continuous and a discrete spectrum of normal modes suggested by the wavevector {right arrow over (k)} is an approximation to the continuous spectrum.

A propagating vector potential solution to the wave equation above that satisfies the periodic boundary conditions is:

$A\left(r,t\right)=\sum _{\stackrel{\rightharpoonup }{k},s}{\stackrel{\rightharpoonup }{e}}_{\stackrel{\rightharpoonup }{k}s}\left(A_{\stackrel{\rightharpoonup }{k}s}{e}^{i\left(\stackrel{\rightharpoonup }{k}·\stackrel{\rightharpoonup }{r}-{\omega }_{k}t\right)}+A_{\stackrel{\rightharpoonup }{k}s}^{*}{e}^{-i\left(\stackrel{\_}{k}·\stackrel{\rightharpoonup }{r}-{\omega }_{k}t\right)}\right)$
where

A_{{right arrow over (k)}s} is a complex amplitude of the electromagnetic radiation;

{right arrow over (e)}_{{right arrow over (k)}s} represents two unit-length polarization vectors; and

m_x, m_y, m_z=0, ±1, ±2, ±3, . . . .

The sum over {right arrow over (k)} represents the sum over the integers (m_x, m_y, m_z), and the sum over s is the sum over the two independent polarizations that are associated with each {right arrow over (k)}. The two polarization vectors are orthogonal as indicated by:
{right arrow over (e)} _{{right arrow over (k)}} ·{right arrow over (e)} _{{right arrow over (k)}s′}=δ_ss′,
and from the gauge condition given above:
{right arrow over (k)}·{right arrow over (e)} _{{right arrow over (k)}s}=0,
for both polarization directions s. The two polarization vectors {right arrow over (e)}_{{right arrow over (k)}1} and {right arrow over (e)}_{{right arrow over (k)}2} form a right-handed coordinate system with a normalized wavevector given by:

${\stackrel{\rightharpoonup }{e}}_{\stackrel{\rightharpoonup }{k}1}\times {\stackrel{\rightharpoonup }{e}}_{\stackrel{\rightharpoonup }{k}2}=\frac{\stackrel{\rightharpoonup }{k}}{\stackrel{\rightharpoonup }{k}}=\stackrel{\rightharpoonup }{\kappa }$
FIG. 2 illustrates a three-dimensional right-handed coordinate system with two independent polarization vectors {right arrow over (e)}_{{right arrow over (k)}s} and a normalized wavevector {right arrow over (κ)} as basis vectors. In FIG. 2, the wavevector {right arrow over (κ)} 202, and the polarization vectors, {right arrow over (e)}_{{right arrow over (k)}1} 204 and {right arrow over (e)}_{{right arrow over (k)}2} 206, define three orthogonal unit length basis vectors of a coordinate system with coordinate axes represented by lines 208, 210, and 212, respectively.

The propagating electric and magnetic field components of the vector potential are:

$\stackrel{\_}{E}\left(\stackrel{\_}{r},t\right)=i\sum _{\stackrel{\_}{k},s}{\omega }_k{\stackrel{\_}{e}}_{\stackrel{\_}{k}s}\left[A_{\stackrel{\_}{k}s}{e}^{i\left(\stackrel{\_}{k}·\stackrel{\_}{r}-{\omega }_{k}t\right)}-A_{\stackrel{\_}{k}s}^{*}{e}^{-i\left(\stackrel{\_}{k}·\stackrel{\_}{r}-{\omega }_{k}t\right)}\right],\mathrm{and}$ $\stackrel{\_}{B}\left(\stackrel{\_}{r},t\right)=\frac{i}{c}\sum _{\stackrel{\_}{k},s}{\omega }_k\left(\stackrel{\_}{\kappa }\times {\stackrel{\_}{e}}_{\stackrel{\_}{k}s}\right)\left[A_{\stackrel{\_}{k}s}{e}^{i\left(\stackrel{\_}{k}·\stackrel{\_}{r}-{\omega }_{k}t\right)}-A_{\stackrel{\_}{k}s}^{*}{e}^{-i\left(\stackrel{\_}{k}·\stackrel{\_}{r}-{\omega }_{k}t\right)}\right]$
Both the electric field {right arrow over (E)}({right arrow over (r)},t) and magnetic field {right arrow over (B)}({right arrow over (r)},t) are propagating wave solutions referred to as the “classical” representation of the electric and magnetic field, are orthogonal to one another, and are both orthogonal to the wavevector {right arrow over (k)}. Note that the wavevector {right arrow over (k)} and the polarization parameter s define what is called a “spatial mode” or “mode” of the electromagnetic radiation field.

FIG. 3 illustrates a representation of electric and magnetic field components of electromagnetic radiation in the right-handed coordinate system shown in FIG. 2. The electromagnetic radiation is directed along the wavevector {right arrow over (κ)} 202 axis. The electric field component {right arrow over (E)}({right arrow over (r)},t) 302 and magnetic field component {right arrow over (B)}({right arrow over (r)},t) 304 are directed along the orthogonal polarization vectors {right arrow over (e)}_{{right arrow over (k)}1} 204 and {right arrow over (e)}_{{right arrow over (k)}2} 206, respectively, and appear frozen at a particular time t.

The energy of the electromagnetic radiation can be determined by evaluating the Hamiltonian:

$\begin{array}{c}H=\frac{1}{2}{\int }_V^{}\left({\mathrm{<figure-callout\; id="0"\; label="\varepsilon "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00002.png"\; state="\{\{state\}\}">\varepsilon </figure-callout>}}_0\stackrel{\rightharpoonup }{E}·\stackrel{\rightharpoonup }{E}+\frac{1}{{\mathrm{<figure-callout\; id="0"\; label="\mu "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00002.png"\; state="\{\{state\}\}">\mu </figure-callout>}}_0}\stackrel{\rightharpoonup }{B}·\stackrel{\rightharpoonup }{B}\right)dV\\ =2{\mathrm{<figure-callout\; id="0"\; label="\varepsilon "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00002.png"\; state="\{\{state\}\}">\varepsilon </figure-callout>}}_{0}V\sum _{\stackrel{\_}{k},s}{\omega }_{\mathrm{<figure-callout\; id="2"\; label="k"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">k</figure-callout>}}^2{A}_{\stackrel{\_}{k}s}{A}_{\stackrel{\_}{k}s}^{*},\end{array}$
where ∈₀ is the electric permittivity of free space;

μ₀ is the magnetic permeability of free space; and

V is the volume of the cavity.

The electric permittivity ∈₀ represents the degree to which a vacuum space can store electrical potential energy under the influence of an electric field, and the magnetic permeability μ₀ represents the degree to which the vacuum modifies the flux of a magnetic field. In a dielectric medium, the electric permittivity is further multiplied by a dielectric constant ∈, which is the degree to which the medium enhances the storage of electrical potential energy, and the magnetic permeability is further multiplied by μ, which is the degree to which the medium further enhances the flux of a magnetic field.

The quantum Hamiltonian operator is given by:

$\hat{H}=\sum _{\stackrel{\_}{k},s}\hslash {\omega }_k\left({\hat{a}}_{\stackrel{\_}{k}s}^{†}{\hat{a}}_{\stackrel{\_}{k}s}+\frac{1}{2}\right),$
where

• • â_{{right arrow over (k)}s} is called the “annihilation operator;”
  • â_{{right arrow over (k)}s} ^† is called the “creation operator;” and
  • â_{{right arrow over (k)}s} ^†â _{{right arrow over (k)}s} is called the “number operator” and is also denoted by {circumflex over (n)}_{{right arrow over (k)}s}.

When the electromagnetic field is quantized, the amplitudes A_{{right arrow over (k)}s} are given by the operators:

${\hat{A}}_{\stackrel{\_}{k}s}=\sqrt{\frac{\mathrm{<figure-callout\; id="2"\; label="ħ"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">\hslash </figure-callout>}}{2{\omega }_k{\mathrm{<figure-callout\; id="0"\; label="\varepsilon "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00002.png"\; state="\{\{state\}\}">\varepsilon </figure-callout>}}_{0}V}}{\hat{a}}_{\stackrel{\_}{k}s},$
which can be substituted in the classical electric and magnetic field equations above to obtain electric and magnetic field operators:

$\hat{E}\left(\stackrel{\_}{r},t\right)=i\sum _{\stackrel{\_}{k},s}\sqrt{\frac{\hslash \omega }{2{\mathrm{<figure-callout\; id="0"\; label="\varepsilon "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00002.png"\; state="\{\{state\}\}">\varepsilon </figure-callout>}}_{0}V}}{e}_{\stackrel{\_}{k}s}[{\hat{a}}_{\stackrel{\_}{k}s}{e}^{i\left(k·r-{\omega }_{k}t\right)}-{\hat{a}}_{\stackrel{\_}{k}s}^{†}{e}^{-i\left(k·r-{\omega }_{k}t\right)}\text{]},\mathrm{and}\hat{B}\left(\stackrel{\_}{r},t\right)=\frac{i}{c}\sum _{\stackrel{\_}{k},s}\left(\stackrel{\rightharpoonup }{\kappa }\times e_{\stackrel{\_}{k}s}\right)\sqrt{\frac{\hslash \omega }{2{\mathrm{<figure-callout\; id="0"\; label="\varepsilon "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00002.png"\; state="\{\{state\}\}">\varepsilon </figure-callout>}}_{0}V}}{e}_{\stackrel{\_}{k}s}[{\hat{a}}_{\stackrel{\_}{k}s}{e}^{i\left(k·r-{\omega }_{k}t\right)}-{\hat{a}}_{\stackrel{\_}{k}s}^{†}{e}^{-i\left(k·r-{\omega }_{k}t\right)}\text{]}$
Both the electric and magnetic field operators are Hermitian and represent measurable electric and magnetic fields.

Because the magnitude of the magnetic field is smaller than the electric field by the factor 1/c, the electric field accounts for most of the interactions with charged matter. As a result, the electric field alone is generally used to characterize the behavior of electromagnetic radiation and any interactions with charged matter, and the magnetic field component can be ignored.

Quantum computation and quantum information processing systems can be performed with a single mode {right arrow over (k)}s of electromagnetic radiation. As a result, the Hamiltonian operator for a single mode of electromagnetic radiation reduces to:

$\hat{H}=\hslash \omega \left({\hat{a}}^{†}\hat{a}+\frac{1}{2}\right),$
where â and ↠replace the mode-dependent operators â_{{right arrow over (k)}k j s j} and â_{{right arrow over (k)} j s j} ^† in the Hamiltonian above. The eigenstates and the corresponding energy eigenvalues of the single mode Hamiltonian are:

$\hat{H}|n\rangle =\hslash \omega \left({\hat{a}}^{†}\hat{a}+\frac{1}{2}\right)\text{|}n\text{〉}=E_n|n\text{〉},$
where |n

is called a “number state,” n is a nonnegative integer called the “photon number” representing the number of photons in an electromagnetic field, and En is an energy eigenvalue or energy of the electromagnetic field.

The annihilation and creation operators operate on a number state as follows:
â|n

=√{square root over (n)}|n−1

â†|n

=√{square root over (n+1)}|n+1

, and
{circumflex over (n)}|n

=n|n

where {circumflex over (n)} represents the operator â^†â and is called the “number operator.” The number states can be generated by repeated application of the annihilation and creation operators to the number states. For example, repeated application of the annihilation operator to a number state lowers the photon number:

$0\rangle =\frac{{\hat{a}}^n}{\sqrt{n!}}\text{}n\text{〉},$
where |0

is called the “vacuum state” and represents the lowest energy state of the electromagnetic radiation. Beginning with the vacuum state, and repeatedly applying the creation operator gives:

$|n\rangle =\frac{{\left({\hat{a}}^{†}\right)}^n}{\sqrt{n!}}\text{|}0\text{〉}$
The number states are orthogonal and form a compete set represented by:

$\langle n^{\prime }|n\rangle ={\delta }_{n^{\prime }n},\mathrm{and}$ $\sum _{n=0}^{\infty }|n\text{〉}\langle n|=1$

In general, the energy eigenvalue equation associated with a number state |n

is:

$\hat{H}|n\rangle =\hslash \omega \left(n+\frac{1}{2}\right)\text{|}n\text{〉}=E_n|n\text{〉}.$
Applying the annihilation and creation operators to the energy eigenvalue equation gives:

$\hat{H}(\hat{a}|n\rangle \text{)}=\hslash \omega \left(n-\frac{1}{2}\right)\text{|}n-1\text{〉}=\left(E_n-\hslash \omega \right)|n-1\text{〉},\mathrm{and}$ $\hat{H}({\hat{a}}^{†}|n\rangle \text{)}=\hslash \omega \left(n+\frac{3}{2}\right)\text{|}n+1\text{〉}=\left(E_n+\hslash \omega \right)|n+1\text{〉},$
which shows that the energy levels of electromagnetic radiation are equally spaced by a quantum of energy

ω. In other words, the excitations of electromagnetic radiation occur in discrete amounts of energy

ω called “photons.” The photon number n refers to the number of photons

ω comprising the electromagnetic radiation.

FIG. 4 is an energy level diagram of a single mode of quantized electromagnetic radiation. Horizontal lines represent energy levels of electromagnetic radiation. Energy level 402 is the lowest energy level, which corresponds to the ground or vacuum state |0

The energy of the vacuum state is

ω/2 or ½ the energy of a single photon

ω. Higher energy levels of electromagnetic radiation are each separated from each other by the same quantum of energy

ω. For example, energy level 404 represents electromagnetic field with a total electromagnetic energy of 5

ω/2, which can be thought of as the energy of electromagnetic radiation having two photons plus the vacuum state energy

ω/2. The annihilation operator a corresponds to removal of a photon from the electromagnetic field, and the creation operator â^† corresponds to addition of a photon to the electromagnetic radiation. For example, the annihilation operator â represents a energy transition from the state |n

410 to the lower energy state |n−1

412. This transition is accomplished by the electromagnetic field giving up a photon to the surroundings. By contrast, the creation operator at represents a transition from the state |n

410 to the higher energy state |n+1

414. This transition is accomplished by the electromagnetic field accepting a photon from the surroundings. Note that typically the surroundings can be an atom, a quantum dot, or any other system that couples to the electromagnetic field through a dipole interaction. Loss or absorption of a photon will involve a simultaneous excitation of the surrounding system and creation or emission of a photon will involve a corresponding de-excitation of the surrounding system.

Photons can be generated by a photon source and transmitted through free space or in an optical fiber. The photon source can generate a single short-duration burst of electromagnetic radiation, called a “pulse,” or generate a sequence or train of pulses, each pulse containing one or more photons that all have the same electromagnetic properties, such as wavelength, phase, and direction. Photons with the same optical properties are called “coherent.” However, the source, a detector, and a medium, such as an optical fiber, separating the source from the detector does not define an optical cavity. The source and the detector are parts of a continuous unidirectional flow of electromagnetic energy with no significant reflection or recycling of the energy. A pulse transmitted through free space or a waveguide can be described by a wavepacket that can be represented by a time-dependent, Gaussian-shaped function given by:

$\xi \left(t\right)={\left(\frac{2{\mathrm{<figure-callout\; id="2"\; label="\Delta "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">\Delta </figure-callout>}}^2}{\pi }\right)}^{1/4}\exp \left\{-i{\omega }_{0}t-{{\mathrm{<figure-callout\; id="2"\; label="\Delta "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">\Delta </figure-callout>}}^2\left(t_0-t\right)}^2\right\},$
where

ω₀ is the central frequency of the pulse spectrum,

t is time,

t₀ is the time at which the peak of the wavepacket is located at a distance z₀ from the photon source, and

Δ² is the variance of the intensity spectrum.

The time t₀ can be determined by z₀/v, where v is the velocity of the pulse traveling through free space or in an optical fiber.

The wavepacket ξ(t) is the amplitude of the pulse, and |ξ(t)|² is a photodetection probability density function of the pulse, where the photodetection probability density function |ξ(t)|² satisfies the normalization condition:

${\int }_{-\infty }^{\infty }dt\xi \left(t\right){|}^2=1$
The probability of photodetection of a photon in the time interval (t₁,t₂) at a distance z₀ from the photon source is given by:

$\mathrm{Probability}\mathrm{of}\left({\mathrm{<figure-callout\; id="1"\; label="t"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">t</figure-callout>}}_1<t_2\right)={\int }_{t_1}^{{\mathrm{<figure-callout\; id="2"\; label="t"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">t</figure-callout>}}_2}dt{\xi \left(t\right)}^2$

FIG. 5 illustrates a probability distribution associated with a pulse output from a source 502 and transmitted in a waveguide 504 to a detector 506. The horizontal line 508 represents the distance z₀ the photon travels from the source 502 to the detector 506, and the horizontal line 510 is a time axis. The curve 512 represents the photodetection probability density function |ξ(t)|². In FIG. 5, the photodetection probability density function |ξ(t)|² 512 is centered at the time t₀, which corresponds to the average time a pulse takes to travel the distance z₀. An area under the curve 512 represents the probability of detecting the pulse within a particular time period. For example, hash-marked region 514 represents the probability of detecting the photon within the time period t₁<t₀<t₂. Time period 516 is called a “time bin” and corresponds to the time period within which the photon is detected at the detector 506.

An Overview of Coherent States

The most common kind of single-mode states are linear superpositions of the number states. There are a number of different possible linear superpositions of the number states, but the coherent state:

$\alpha \rangle =\exp \left(-\frac{1}{2}{\alpha }^2\right)\sum _{n=0}^{\infty }\frac{{\alpha }^n}{\sqrt{n!}}n\rangle$
is a linear superposition of the number states used in many applications of quantized electromagnetic radiation. The coherent states are eigenstates of the annihilation operator:
â|α

=α|α

where taking the complex conjugate gives:

α|{circumflex over (a)}^†=

α|α*
However, the coherent state |α

is not an eigenstate of the creation operator {circumflex over (a)}^† because the summation over a cannot be rearranged to give the coherent state from â^†|α

The coherent state expectation value for the number operator:

n

=

α|{circumflex over (n)}|α

=|α| ²
indicates that |α|² is the mean number of photons. The probability of detecting n photons in a measurement of the number of photons is a Poisson distribution:

$P_n={\langle n|\alpha \rangle }^2=\exp \left(-{\alpha }^2\right)\frac{{\alpha }^2}{n!}$
The Poisson distribution approaches a Gaussian distribution for large values of |α|².

The coherent state is a quantum state whose properties most closely resemble a classical electromagnetic wave of stable amplitude and fixed phase. For example, the electric field operator corresponding to an electric field propagating in the z direction, with the mode subscripts k and s removed, is:

$\begin{array}{c}\hat{E}\left(\Omega \right)=\frac{1}{2}\left(\hat{a}{e}^{-\mathrm{i\Omega }}+{\hat{a}}^{†}{e}^{\mathrm{i\Omega }}\right)\\ =\hat{X}\cos \Omega +\hat{Y}\sin \Omega \end{array}$
where the time t and displacement z are contained in the phase angle:

$\Omega \left(z,t\right)=\omega t-\mathrm{kz}-\frac{\mathrm{<figure-callout\; id="2"\; label="\pi "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">\pi </figure-callout>}}{2}$
and the electric field is measured in units of √{square root over (

ω/2∈₀V)}.

The coherent state is a nearly classical state because it gives the correct sinusoidal form for the electric field expectation value or coherent signal:

$\begin{array}{c}\sum =\langle \alpha \hat{E}\left(\Omega \right)\alpha \rangle \\ =\alpha \cos \left(\Omega -\varphi \right)\end{array}$
where α=|α|e^iφ, and

φ is the mean phase angle of the coherent state excitation of the mode.

Polarization States and Stokes Parameters

In this subsection, polarization states of electromagnetic radiation are discussed. As described above with reference to FIG. 3, electromagnetic radiation may be treated as propagating transverse electromagnetic waves. It is the electric field component, alone, that can be used to represent an electromagnetic wave, because the electric field accounts for most of the interactions with charged matter and the magnitude of the magnetic field is smaller than the electric field by the factor 1/c. As shown in FIG. 3, when both the oscillating electric field {right arrow over (E)}({right arrow over (r)},t) component and the associated wavevector {right arrow over (k)} of an electromagnetic field reside in a plane-of-vibration the field is said to be “linearly polarized.” Definite polarization states can be created by transmitting electromagnetic radiation comprising numerous randomly polarized electromagnetic waves through one or more polarizers. Each polarizer is a device that transmits only the electromagnetic waves with electric field components that are aligned with the polarization axis of the polarizer.

Any two orthogonal linear polarization states can be used to define a polarization basis, denoted by {|H

,|V

. The first polarization state |H

represents an electromagnetic wave polarized in a first direction called the “horizontal polarization,” and the second polarization state |V

represents an electromagnetic wave polarized in a second direction orthogonal to the first direction and called “vertical polarization.” The polarization basis states satisfy the following conditions:

H|H

=

V|V

=1, and

H|V

=1
FIGS. 6A-6B show plots of the polarization basis states |H

and |V

In FIGS. 6A-6B, mutually perpendicular axes, such as mutually perpendicular axes 601-603 in FIG. 6A, represent the x, y, and z Cartesian coordinate axes, respectively. FIG. 6A shows a vertical polarization state |V

of an electric field {right arrow over (E)}_y(z,t) 604 that lies in the yz-plane. Directional arrow 606 represents the direction the electric field {right arrow over (E)}_y(z,t) 604 propagates toward a plane of observation 608. From the plane of observation 608, one can observe the electric field {right arrow over (E)}_y(z,t) 604 progress through one complete oscillatory cycle as the wave propagates along the z-axis through one wavelength λ. The oscillatory cycle is represented by a double-headed directional arrow 610. FIG. 6B shows a horizontal polarization state |H

of an electric field {right arrow over (E)}_x(z,t) 612 that lies in the xz-plane. The associated horizontal oscillatory cycle is represented by a double-headed directional arrow 616 in the plane of observation 608.

The polarization basis {|H

|V

can also be used to construct an infinite number of polarization states represented by |χ

These polarization states can be represented mathematically as a coherent linear superposition of states:

$\chi \rangle =\cos \left(\frac{\theta }{2}\right)H\rangle +e^{\mathrm{i\varphi }}\sin \left(\frac{\theta }{2}\right)V\rangle$
where

0≦θ<π, and

0≦φ<2π.

An infinite number of polarization states of an electromagnetic wave can be geometrically represented by a three-dimensional Bloch sphere, which in this case is also called the “Poincare sphere.”

FIG. 7 illustrates a Poincare sphere representation of polarization states. As shown in FIG. 7, lines 701-703 are orthogonal coordinate axes, respectively, and a Poincare sphere 704 is centered at the origin. There are an infinite number of points on the Poincare sphere 704, each point representing a unique pure polarization state |χ

of an electromagnetic wave. For example, a point 705 represents a polarization state |χ

that simultaneously comprises, in part, the state |H

and, in part, the state |V

The six points 706-711 identify intersections between the Poincare sphere 704 and the coordinate axes 701-703. The points 706 and 707 identify the polarization basis states |H

and |V

respectively, and the points 708-711 represent orthogonal polarization states:

${45}^0\rangle =\frac{1}{\sqrt{2}}\left(H\rangle +V\rangle \right),-{45}^0\rangle =\frac{1}{\sqrt{2}}\left(H\rangle -V\rangle \right),R\rangle =\frac{1}{\sqrt{2}}\left(H\rangle +iV\rangle \right),\mathrm{and}$ $L\rangle =\frac{1}{\sqrt{2}}\left(H\rangle -iV\rangle \right),$
respectively.

FIGS. 8A-8D show plots of the four polarization states |45°

|−45°

|R

and |L

respectively. FIG. 8A shows the 45° polarization state |45°

that lies within a plane-of-vibration 802 tilted at a 45° angle to the horizontal xz-plane. The oscillatory cycle of the polarization state |45°

is represented by a double-headed directional arrow 804. FIG. 8B shows the −45° polarization state |45°

that lies within a plane-of-vibration 806 tilted at a −45° angle to the horizontal xz-plane. The oscillatory cycle of the polarization state |45°

is represented by a double-headed directional arrow 808. FIG. 8C shows a right-handed circularly polarization state |R

comprising the vertical and horizontal polarized fields 604 and 612, shown in FIGS. 6A and 6B, with a relative phase difference δ of −π/2. The oscillatory cycle of the right-handed polarization state is represented by orthogonal double-headed directional arrows 810 and 812 that appear to rotate in a clockwise direction 814. FIG. 8D shows a left-handed circularly polarization state also comprising the vertical and horizontal polarized fields 604 and 612 with a relative phase difference δ of π/2. The oscillatory cycle of the left-hand polarization state is represented by orthogonal double-headed directional arrows 816 and 818 that appear to rotate in a counterclockwise direction 820.

Any polarization state can be represented by a linear combination of four quantities, called “Stokes parameters.” Stokes parameters are a convenient way to represent the polarization states of quasi-monochromatic electromagnetic radiation because electromagnetic radiation measurements typically can only determine intensities or the number of photons and not the polarization states. The Stokes parameters all have the same dimension and, for a monochromatic wave, are given by the four quantities:
S ₀ =

a ₁ ²

+

a ₂ ²

S ₁ =

a ₁ ²

a ₂ ²

S₂=2

a₁a₂ cos δ

and
S₃=2

a₁a₂ sin δ

where

the symbol “

•

” represents the average value;

a₁ and a₂ are instantaneous amplitudes of two different orthogonal components E_x and E_y of the electric field vector; and

δ is the phase difference between the components E_x and E_y.

Only three of the Stokes parameters for a monochromatic wave are independent since the parameters are also related by the identity:
S ₀ ² =S ₁ ² +S ₂ ² +S ₃ ²
Note that for a partially coherent quasi-monochromatic wave, the Stokes parameters are related by the inequality:
S ₀ ² ≧S ₁ ² +S ₂ ² +S ₃ ²
The Stokes parameters are related to one another by the following Stokes relations:
S₁=S₀ cos 2χ cos 2ψ,
S₂=S₀ cos 2χ sin 2ψ, and
S₃=S₀ sin 2χ
where

$0\le \psi \le \pi ,\mathrm{and}-\frac{\mathrm{<figure-callout\; id="4"\; label="\pi "\; filenames="US07849122-20101207-D00005.png"\; state="\{\{state\}\}">\pi </figure-callout>}}{4}\le \chi \le \frac{\mathrm{<figure-callout\; id="4"\; label="\pi "\; filenames="US07849122-20101207-D00005.png"\; state="\{\{state\}\}">\pi </figure-callout>}}{4}$

FIG. 9 illustrates a geometric representation of the Stokes parameters S₁, S₂, and S₃. As shown in FIG. 9, lines 902-904 are orthogonal x, y, and z Cartesian coordinate axes, respectively. The sphere 906 of radius S₀ is a geometric representation of all the different polarization states. The Stokes parameters S₁, S₂, and S₃ are regarded as the Cartesian coordinates of a point P 908 on the sphere 906 and 2χ and 2ψ are the spherical angular coordinates. For every possible polarization state of a given intensity S₀, there is a corresponding point on the sphere 906 and vice versa. Right-handed polarization is represented by points on the sphere 906 that lie above the equatorial xy-plane 910, and left-handed polarization is represented by points that on the sphere 906 that lie below the equatorial xy-plane 910. For linearly polarized electromagnetic radiation, the phase difference δ is zero or an integral multiple of π and the parameter S₃ is zero. In other words, linearly polarized electromagnetic waves are represented by points that lie at the intersection of the sphere 906 and the xy-plane 910. For circularly polarized electromagnetic radiation,

a₁

equals

a₂

and the phase difference δ is π/2 or −π/2. Therefore, right-handed circularly polarized electromagnetic radiation is represented by a point 912, and left-handed circularly polarized electromagnetic radiation is represented by a point 914. Note that for partially coherent quasi-monochromatic waves, the states are represented by points that lie inside the sphere 906, as indicated by the inequality above.

Typically, the Stokes parameters are normalized by dividing each parameter by the parameter S₀, which is equivalent to using an incident beam of unit intensity. The Stokes parameters (S_O,S₁,S₂,S₃) for randomly polarized electromagnetic radiation in the normalized representation are (1, 0, 0, 0), which corresponds to the center of the sphere 906. The normalized Stokes parameters are listed in Table I:

	TABLE I
	|H	|V	|450	|−450	|R	|L

	S 0	1	1	1	1	1	1
	S 1	1	−1	0	0	0	0
	S 2	0	0	1	−1	0	0
	S 3	0	0	0	0	1	−1

The Stokes parameters of any quasi-monochromatic wave of electromagnetic radiation may be determine by intensity or photon number measurements and are given by the relations:

${\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00002.png"\; state="\{\{state\}\}">S</figure-callout>}}_0=I\left(0°,0\right)+I\left(90°,0\right),{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1=I\left(0°,0\right)-I\left(90°,0\right),{\mathrm{<figure-callout\; id="2"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">S</figure-callout>}}_2=I\left(45°,0\right)-I\left(-45°,0\right),\mathrm{and}$ $S_3=I\left(45°,\frac{\pi }{2}\right)-I\left(-45°,\frac{\pi }{2}\right),$
where I(θ,∈) represents intensity of electromagnetic radiation with the electric field vibrations making an angle of θ with the x-axis when the y-component is subjected to a retardation ∈ with respect to the x-component. For example, the intensities I(0°, 0) and I(90°, 0) represent the intensities of horizontally and vertically polarized electromagnetic radiation, I(45°, 0) and I(−45°, 0) can represent the intensity of 45° and −45° polarized electromagnetic radiation, and

$I\left(45°,\frac{\pi }{2}\right)\mathrm{and}I\left(-45°,\frac{\pi }{2}\right)$
represent right-handed and left-handed circularly polarized electromagnetic radiation.

Note that the parameter S₀ represents the total intensity. The parameter S₁ equals the excess in intensity of electromagnetic radiation transmitted by a polarizer which accepts linear polarization with an angle of θ equal to 0° over electromagnetic radiation transmitted with by a polarizer which accepts linear polarization with an angle of θ equal to 90°. The parameter S₂ has a similar interpretation. The parameter S₃ equals the excess in intensity of electromagnetic radiation transmitted by a polarizer which accepts right-handed circularly polarized electromagnetic radiation, over left-handed circularly polarized electromagnetic radiation.

EMBODIMENTS OF THE PRESENT INVENTION

Various system embodiments of the present invention are directed to self-authenticating QRBGs that can be integrated into optoelectronic devices. Method embodiments of the present invention include tomographic analysis, which is used to evaluate and authenticate the randomness of a sequence of random bits generated by the system embodiments of the present invention.

I. Quantum Random Bit Generators

FIG. 10 illustrates a general schematic representation of a QRBG 1000 that represents an embodiment of the present invention. The QRBG 1000 comprises a state generator 1002, a polarization states analyzer (“PSA”) 1004, a raw bit generator (“RBG”) 1006, and a system control 1008. The state generator 1002 outputs a 45° polarized pulse 1010 of electromagnetic radiation in a coherent state:

${\mathrm{<figure-callout\; id="45"\; label="\alpha "\; filenames="US07849122-20101207-D00000.png,US07849122-20101207-D00005.png"\; state="\{\{state\}\}">\alpha </figure-callout>}}_{45}\rangle =\frac{1}{\sqrt{2}}\left({\alpha }_H\rangle +{\alpha }_V\rangle \right)$
where

|α_H

represents a horizontally polarized coherent state, and

|α_V

represents a vertically polarized coherent state.

The term “horizontal” refers to electromagnetic waves with the electric field component polarized in the plane of the QRBG 1000, and the term “vertical” refers to electromagnetic waves with the electric field component polarized orthogonal to the plane of the QRBG 1000. The polarized pulse of electromagnetic radiation |α₄₅

1010 is split so that a first portion of the pulse |α₄₅

1010 is transmitted to the PSA 1004 and a second portion of the pulse |α₄₅

1010 is transmitted to the RBG 1006. Various system embodiments for the PSA 1004 and the RBG 1006 are described below with reference to FIGS. 11-14. For each pulse |α₄₅

1010 generated by the state generator 1002, the PSA 1004 projects the initial state onto four different polarized coherent states of electromagnetic radiation. The four differently polarized coherent states are: (1) a 45° polarized pulse |α₄₅

1012, (2) a −45° polarized pulse |α₋₄₅

1013, (3) a right-handed circularly polarized pulse |α_R

1014, and (4) a left-handed circularly polarized pulse |α_L

1015. The PSA 1004 includes a detection system 1018 that detects the states 1012-1015 and transmits the detection results to the system control 1008. The RBG 1006 may include an optional attenuator 1020 that reduces the number of photons in the pulse |α₄₅

1010 as follows:

${\mathrm{<figure-callout\; id="45"\; label="\alpha "\; filenames="US07849122-20101207-D00000.png,US07849122-20101207-D00005.png"\; state="\{\{state\}\}">\alpha </figure-callout>}}_{45}\rangle \stackrel{\mathrm{ATT}}{\to }\{\begin{array}{cc}0\rangle & \mathrm{when}\mathrm{no}\mathrm{pulse}\mathrm{is}\mathrm{present}\\ \frac{1}{\sqrt{2}}\left(H\rangle +V\rangle \right)& \mathrm{when}a\mathrm{pulse}\mathrm{is}\mathrm{present}\end{array}$
where

|H

represents a horizontally polarized pulse comprising a single photon; and

|V

represents a vertically polarized pulse comprising a single photon.

The RBG 1006 includes a detection system 1022 that detects the presence of a single photon and transmits the detection result to the system control 1008.

The photon polarization states |H

and |V

are used to encode bits. For example, detection of the state |H

can be used to represent the binary number “1” and detection of the state |V

can be used to represent the binary number “0.” For N operation cycles of the QRBG 1000, the system control 1008 receives the detection results supplied by the PSA 1004 and the RBG 1006 and outputs a sequence of random bits represented by a vector {right arrow over (s)}_m, where m represents the number of random bits and m<N. Method embodiments described below with reference to FIGS. 15-22 are directed to constructing the sequence of random bits {right arrow over (s)}_m.

FIGS. 11-16 illustrate various QRBG system embodiments of the present invention that are in accordance with the general schematic QRBG 1000 described above. In the interest of brevity, components that are common to the various QRBGs, shown in FIGS. 11-16, are provided with the same reference numerals and an explanation of their structure and function is not repeated.

FIG. 11 illustrates a schematic representation of a first QRBG 1100 that represents an embodiment of the present invention. The QRBG 1100 comprises an electromagnetic radiation transmission layer 1102 and a system control 1104. The transmission layer 1102 includes a laser diode 1106 and a ridge waveguide 1108 that splits into three waveguides 1110-1112. The transmission layer 1102 can be fabricated from a slab of silicon oxynitride (“SiON”) or a suitable optical polymer, such as SU-8. The waveguides 1108 and 1110-1112 can be formed in the slab using various well-known reactive ion etching and lithographic methods. The QRBG 1100 includes three half-wave polarization rotators 1114-1116, and a quarter-wave polarization rotator 1117. The kinds of polarization rotators that can be used to introduce polarization into the waveguides 1110 and 1112 are described below with reference to FIGS. 12 and 13. The QRBG 1100 includes three polarizing beamsplitters 1124-1126 located near the terminus of the waveguides 1110-1112, respectively. The polarizing beamsplitters 1124-1126 are each comprised of a tapered waveguide located in close proximity to the waveguides 1110-1112, respectively. For example, polarizing beamsplitter 1124 includes a shorter waveguide 1136. The tapered portions of the shorter waveguides are each configured so that vertically polarized pulse are evanescently extracted from the waveguides 1110-1112 while horizontally polarized pulses continue in the waveguides 1110-1112. The system control 1104 includes p-i-n photodetectors 1128-1131 and two avalanche photodiodes 1132 and 1133.

FIGS. 12A-12B illustrate a slotted polarization rotator 1200 that represents an embodiment of the present invention. FIG. 12A illustrates a top view of the slotted polarization rotator comprising 8 slots represented by shaded rectangles, such as shaded rectangle 1202. The slots have a substantially periodic spacing p and length L. FIG. 12B illustrates a cross-sectional view of the slotted polarization rotator 1200 shown in FIG. 12A that represents an embodiment of the present invention. The slots are etched into the ridge waveguide 1204 at an angle s and to a depth d. The slotted polarization rotator 1200 can be fabricated to operate similar to waveplate, which are well-known in larger scale optical applications. For example, in various embodiments of the present invention, the slotted polarization rotator 1200 can be fabricated with appropriately selected parameters L, p, φ, and d so that the slotted polarization rotator 1200 can be operated similar to a half-wave plate or a quarter-wave plate. For a more detailed description of slotted polarization rotators see “Compact polarization converter in InP-based material,” by Maria V. Kotlyar et al., Optics Express, Vol. 13, No. 13, June 2005.

FIGS. 13A-13B illustrate a slanted-ridge polarization rotator 1300 that represents an embodiment of the present invention. FIG. 13A illustrates an isometric view of the slanted-ridge polarization rotator 1300 that represents an embodiment of the present invention. The slanted-ridge polarization rotator 1300 is of length L, is embedded in a ridge waveguide 1302, and comprises a sloped sidewall 1304 and an opposing vertical sidewall 1306. The slanted-ridge polarization rotator 1300 is supported by a suitable insulating layer 1308, such as SiO₂. FIG. 13B illustrates a cross-sectional view of the slanted-ridge polarization rotator 1300 that represents an embodiment of the present invention. The slanted-ridge polarization rotator 1300 can also be fabricated to operate similar to a waveplate. For example, in various embodiments of the present invention, the slanted-ridge polarization rotator 1300 can be fabricated with appropriately selected heights H, h, h_r, width W, and length L so that the slanted-ridge polarization rotator 1300 can be operated similar to a half-wave plate or a quarter-wave plate. For further description of slanted-ridge waveguides see “Passive silicon-on-insulator polarization-rotating waveguides,” Optical Engineering 45(4), 044603, April 2006; “Bending characteristics of asymmetric SOI polarization rotators,” IEEE Photonic Technology Letters, Vol. 17, No. 10, October 2005; “Slanted-rib waveguide InGaAsP-InP polarization converters,” J. of Lightwave Technology, Vol. 22, No. 5, May 2004.

Returning to FIG. 11, the following is a general description of the operation of the QRBG 1100 for a single pulse generated by the laser diode 1106 that represents an embodiment of the present invention. The laser diode 1106 can be configured to emit coherent pulses of electromagnetic radiation in either a horizontal polarization state |α_H

or a vertical polarization state |α_V

The half-wave polarizing rotator 1114 receives the coherent pulse and outputs a 45° polarized coherent pulse |α₄₅

The pulse |α₄₅

is transmitted through a 3-way junction 1140 to give:
|α₄₅

→c _A|α₄₅

_A +c _B|α₄₅

_B +c _C|α₄₅

_C
where

|c_A|²+|c_B|²+|c_C|²=1,

|α₄₅

_A represents a 45° polarized pulse transmitted in the waveguide 1110,

|α₄₅

_B represents a 45° polarized pulse transmitted in the waveguide 1111, and

|α₄₅

_C represents a 45° polarized pulse transmitted in the waveguide 1112.

The waveguide 1111 may optionally include an attenuator 1142 that reduces the intensity of the pulse |α₄₅

_B. The optional attenuator 1142 and polarizing beamsplitter 1125 corresponds to the RBG 1006, described above with reference to FIG. 10, and can be used to construct a sequence of bits by assigning the binary number “0” to a detection event at one of the avalanche photodiodes and assigning the binary number “1” to a detection event at the other avalanche photodiode. In various embodiments of the present invention, the attenuator 1142 can be fabricated by implanting a dopant in the waveguide 1111 that reduces the intensity of the pulse |α₄₅

_B to a single photon of energy represented by:

${{\mathrm{<figure-callout\; id="45"\; label="\alpha "\; filenames="US07849122-20101207-D00000.png,US07849122-20101207-D00005.png"\; state="\{\{state\}\}">\alpha </figure-callout>}}_{45}\rangle }_B\stackrel{\mathrm{ATT}}{\to }45°\rangle$
The polarizing beamsplitter 1125 splits the pulse |45°

so that there is a substantially equal probability of detecting a single photon pulse at the avalanche photodiode 1132 or detecting a single photon pulse at the avalanche photodiode 1133, which can be represented in the {|H

|V

basis by:

$45°\rangle \stackrel{\mathrm{BS}}{\to }\frac{1}{\sqrt{2}}\left(H\rangle +V\rangle \right)$
where

|H

represents a horizontally polarized single photon pulse transmitted in the waveguide 1111 and detected by the avalanche photodiode 11132, and

|V

represents a vertically polarized single photon pulse transmitted in the waveguide 1136 and detected by the avalanche photodiode 1133.

The pulses |α₄₅

_A and |α₄₅

_C transmitted in the waveguides 1110 and 1112, respectively, are used to perform tomographic analysis as described below with reference to FIGS. 15-22. The waveguides 1110 and 1112 and corresponding polarization rotators 1115-1117 correspond to the PSA 1004, described above with reference to FIG. 10. The quarter-wave polarization rotator 1117 can be configured to delay the vertical polarization component of the incoming pulse by ¼ of a wavelength, and the half- wave polarization rotators 1115 and 1116 can be configured to rotate the polarization by 45°.

The following is a description of how the polarization rotators 1115 and 1117 operate on a pulses with an arbitrary polarization transmitted in the waveguide 1110. Before the pulse reaches the quarter-wave polarization rotator 1117, the pulse is represented in the {|α_H

α_V

basis by:
|β

_A =a|α _H

_A +b|α _V

_A
with |a|²+|b|²=1. The operation performed by the quarter-wave polarization rotator 1117 is characterized by:

${\beta \rangle }_A\stackrel{\mathrm{QWPR}}{\to }a{{\alpha }_H\rangle }_A+\mathrm{ib}{{\alpha }_V\rangle }_A$
Following the quarter-wave polarization rotator 1117 by appropriately fabricated half-wave polarization rotator 1115 gives:

$a{{\alpha }_H\rangle }_A+\mathrm{ib}{{\alpha }_V\rangle }_A\stackrel{\mathrm{HWPR}}{\to }\frac{1}{\sqrt{2}}\left[\left(a+\mathrm{ib}\right){{}_H\rangle }_A+\left(-a+\mathrm{ib}\right){{\alpha }_V\rangle }_A\right]$
The polarizing beamsplitter 1124 is configured so that the state |α_H

_A is transmitted to the p-i-n photodetector 1128, and the state |α_V

_A is transmitted to the p-i-n photodetector 1129.

Note that when the is initially prepared in a state with a=b=1/√{square root over (2)}, the pulse just before reaching the quarter-wave polarization rotator 1117 is in the state |α₄₅

_A. The pulse emerging from the quarter-wave polarization rotator 1117 is right circularly polarized and represented by |α_R

_A, which, followed by the half-wave polarization rotator 1115, projects the pulse onto the state:

$\frac{1}{2}\left[\left(1+i\right){{\alpha }_H\rangle }_A+\left(-1+i\right){{\alpha }_V\rangle }_A\right]$
Hence the probability of detecting the horizontally polarized pulse |α_H

_A at the p-i-n photodetector 1128 is ½, and the probability of detecting the vertically polarized pulse |α_V

_A at the p-i-n photodetector 1129 is ½.

In general, a pulse with an arbitrary polarization transmitted in the waveguide 1112 just before the half-wave polarization rotator 1116 in the {|α_H

α_V

basis is characterized by:
|β

_C =a|α _H

_C +b|α _V

_C
with |a|²+|b|²=1. The operation performed by the half-wave polarization rotator 1116 is characterized by:

$a{{\alpha }_H\rangle }_C+b{{\alpha }_V\rangle }_C\stackrel{\mathrm{HWPR}}{\to }\frac{1}{\sqrt{2}}\left[\left(a+b\right){{}_H\rangle }_C+\left(-a+b\right){{\alpha }_V\rangle }_C\right]$

Note that when the pulse is initially prepared in a state with a=b=1/√{square root over (2)}, the pulse just before reaching the half-wave polarization rotator 1116 is in the state |α₄₅

_C. The state of the pulse emerging from the half-wave polarization rotator 1116 is give by:
|α_H

_C
In other words, the half-wave polarization rotator 1116 projects the state |α₄₅

_C onto the state |α_H

_C. When initially a=1/√{square root over (2)} and b=−1/√{square root over (2)}, the pulse just before reaching the half-wave polarization rotator 1116 is in the state |α₋₄₅

_C. The state of the pulse emerging from the half-wave polarization rotator 1116 is give by:
|α_V

_C
In other words, the half-wave polarization rotator 1116 projects the state |α₄₅

_C onto the state |α_V

_C.

Although the present invention has been described in terms of a particular system embodiment QRBG 1100, it is not intended that the invention be limited to this embodiment. Modifications within the spirit of the invention will be apparent to those skilled in the art. For example, in another embodiment of the present invention, those skilled in the art would recognize that rather than using a single 3-way junction 1140, shown in FIG. 11, two Y-junctions can be used to split the pulse |α₄₅

As shown in FIG. 14A, Y- junctions 1402 and 1404 split the pulse |α₄₅

into the pulses |α₄₅

_A, |α₄₅

_B, and |α₄₅

_C. In another embodiment of the present invention, rather than using the laser diode 1106, shown in FIG. 11, to generate a pulse, a light-emitting diode (“LED”) 1406 and a polarizing beamsplitter 1408 can be used to generate the pulse |α₄₅

as shown in FIG. 14B. In other embodiments of the present invention, rather than using the transmission layer 1102 with ridge waveguides 1108, 1110-1112 to direct the propagation of pulses, the transmission layer 1102 can be replaced by a photonic crystal with corresponding photonic crystal waveguides that can be used to direct the propagation of pulses. In other embodiments of the present invention, the transmission layer 1102 and ridge waveguides 1108, 1110-1112 can be replaced with optical fibers, the polarization rotators 1114-1118 can be replaced with half- and quarter-wave plates, and the polarizing beamsplitters can be replaced with polarization couplers. For example, FIG. 14C illustrates a schematic representation of a QRBG 1410 that represents an embodiment of the present invention. The QRBG 1410 comprises a laser diode 1412, an optical fiber 1414, a waveguide coupler 1416 that splits the optical fiber 1414 into three optical fibers 1418-1420. The QRBG 1410 also includes three half-wave plates 1422-1424, a quarter-wave plate 1426, and three polarizing couplers 1428-1430. In another embodiment of the present invention, the laser diode 1412 can be replaced with an LED and a half-wave plate.

II. Tomographic Analysis

The detection events at the avalanche photodiodes 1132 and 1133 can be recorded by the system control 1104 are used construct a sequence of random bits. The system control 1104 may record four different detection events. For example, when a pulse is detected by the avalanche photodiode 1132, the binary number “1” is recorded, and when a pulse is detected by the avalanche photodiode 1133, the binary number “0” is recorded. When no pulse is detected at either of the avalance photodiodes 1132 and 1133, “No Pulse” is recorded, and when pulses are detected at both of the avalance photodiodes 1132 and 1133, an “Error” is recorded. Table I summarizes the four kinds of events that may be recorded by the system control 1104 for each pulse |α₄₅

output from the state generator 1002, shown in FIG. 10:

TABLE II
Detection at	Photon State	Information Recorded
APD 1132	|H	“1”
APD 1133	|V	“0”
APD 1132 and APD 1133	|H  and |V	Error
No pulse detected	|0	No Pulse

Each detection event recorded by the system control 1104 is called a “raw count.” FIG. 15 illustrates a hypothetical sequence of N raw counts 1500 recorded by the system control 1104 for a sequence of N pulses |α₄₅

generated by the laser diode 1106 that represents an embodiment of the present invention. The sequence of N raw counts 1500 includes sequences of binary numbers “0” and “1” separated by a raw count “No Pulse” 1502 and a raw count “Error” 1504. Method embodiments of the present invention, described below with reference to FIGS. 17-22, are directed to using the polarization states |α_R

_A, |α_L

_A, |α₄₅

_C, and |α₋₄₅

_C to sift the sequence of N raw counts and obtain the sequence of random bits {right arrow over (s)}_m. Note that detection of pulses at the detectors 1128-1133 are synchronized.

FIG. 16 illustrates generating a hypothetical sequence of random bits from a sequence of raw counts that have been generated by the QRBGs 1100 shown in FIG. 11. The QRBG 1100 generates the sequence of N raw counts 1500, shown in FIG. 15. The “No Pulse” and “Error” detection events recorded in the sequence of N raw counts 1500, such as “No Pulse” 1502 and the “Error” 1504, are removed from to produce a raw sequence of n bits represented by a column vector {right arrow over (r)}_n 1602, where n<N. The raw sequence 1602 includes a number of hypothetically biased bits, such as hypothetically biased bits 1504-1506. The bias may be the result of hardware imperfections or, in a worst cased scenario, the state generator 1002, such as the laser diode 1106, is controlled by a entity that wishes to generate a biased sequence of bits. Quantum tomographic analysis methods of the present invention include constructing an m×n Toeplitz matrix T_m×n, which is used to sift out the biased bits in the raw sequence {right arrow over (r)}_n 1602 using matrix multiplication as follows:
{right arrow over (s)} _m =T _m×n {right arrow over (r)} _n
where m<n<N. The sequence of random bits {right arrow over (s)}_m is represented by a column vector 1608.

“Cryptographic Hardware and Embedded Systems CHES 2003,” by C. D. Walter et al. (Eds.) pp. 166-180, Springer-Verlag (2003), and in particular the chapter “True Random Number Generators Secure in a Changing Environment,” by Barak et al., provides a mathematical definition of the Toeplitz matrix. The following discussion provides the insight needed to construct the Toeplitz matrix in accordance with the Barak et al. reference.

In order to emphasize that the methods of the present invention can be used to generate a sequence of true random numbers in spite of biases in the states output from the state generator 1002, the quantum mechanical-based method embodiments of the present invention are described below with reference to generating a sequence of random bits {right arrow over (s)}_m using an adversarial scenario. FIG. 17 illustrates the QRBG 1000 described above with reference to FIG. 10 under an adversarial scenario. In the adversarial scenario, the state generator 1002 falls under the control of an adversary, called “Eve” 1702, and the rest of the QRBG 1000 is under the control of a user, called “Alice” 1704. Eve wants to generate a sequence of bits that appears random to Alice 1704, and is partially known to Eve. Because Alice generates the random bits using only the states |H

and |V

, the following analysis is restricted to a subspace spanned by the basis {|H

, |V

. Therefore it assumed that Eve generates coherent states of the form:
|ψ

=c|H

+d|V

where

|c|²+|d|²=1,

0≦|c|²≦1, and

0≦|d|²≦1

Eve is assumed to have no knowledge of the measurements Alice applies to photons in the state |ψ

When Eve prepares pulses, all in the same pure state |ψ

Alice can perform a measurement on each pulse and obtain the density matrix:

$\hat{\rho }=\left[\begin{array}{cc}{c}^2& c{d}^{*}\\ c^{*}d& {d}^2\end{array}\right]$
The density matrix {circumflex over (ρ)} represents the maximum amount of information that Alice can obtain about the state of the pulses Alice is provided by Eve. Alice can determine the elements of the density matrix {circumflex over (ρ)} by performing tomographic analysis on pulses provided by Eve. The tomographic analysis, also called “self-authentication,” is used to evaluate randomness of a sequence of bits. Tomographic analysis of quantum states is well-known in the art and is described, for example, in a reference by James et al., “Measurement of Qubits,” Phys. Rev. A, Vol. 64, 052312. Tomographic analysis is used to identify the states |ψ

prepared by Eve. As described in the James et al. reference, in order to perform tomographic analysis on a b-qubit system, (4^b−1) different expectation values are needed to determine the associated density matrix {circumflex over (ρ)}. As a result, numerous copies of identical states are needed for measurements of the expectation values. The (4^b−1) different expectation values and the normalization requirement for the states ideally generates 4^b independent restrictions on 2^b complex coefficients of a general b-qubit system, permitting an analytic solution for the density matrix {circumflex over (ρ)} and 2^b complex coefficients defining the measured state.

Eve may also attempt to bias a sequence in a way that is known to her but appears random to Alice by transmitting pulses in a statistical mixture of the states |ψ_i

(=c_i|H

+d_i|V

), the state of each pulse having an associated probability p_i. Alice performs tomographic analysis to determine the density matrix operator:

$\hat{\rho }=\sum _i{p}_i{\psi }_i\rangle \langle {\psi }_i$
and the associated density matrix:

$\hat{\rho }=\left[\begin{array}{cc}\sum _i{p}_i{c_i}^2& \sum _i{p}_i{c}_i{d}_i^{*}\\ \sum _i{p}_i{c}_i^{*}{d}_i& \sum _i{p}_i{d_i}^2\end{array}\right]$
where

$\sum _i{p}_i{c_i}^2={\mathrm{Pr}}_H$
is the probability of measuring the state |H

and

$\sum _i{p}_i{d_i}^2={\mathrm{Pr}}_V$
is the probability of measuring the state |V

The density matrix operator and the density matrix are compositions of pure state density matrix operators and associated density matrices. Note that although Eve prepares and knows the state |ψ_i

of each pulse that Alice is measuring each time, Eve cannot control the outcome of Alice's measurement on a state |ψ_i

because the outcome of each measurement performed by Alice is governed by the laws of quantum mechanics.

Alice performs the tomographic analysis to determine the density matrix {circumflex over (ρ)} and to evaluate the quality of a source of randomness. The quality of a source of randomness can be strictly evaluated using a minimum entropy (“min-entropy”) function defined as follows:

$H_{\min }\left(X\right)\equiv -{\log }_2\left(\underset{x\in X}{\max }\mathrm{Pr}\left(x\right)\right)$
where

X is a random variable;

Pr(x) is the probability of a event x; and

$\underset{x\in X}{\max }\mathrm{Pr}\left(x\right)$
means the maximum probability Pr(x) over every event x in X.
In other words, the min-entropy can be thought of as a measure of the amount of randomness in a probability distribution ranging from “0” to “1,” where “0” means an event occurs with certainty or not at all, and “1” means whether or not an event occurs is truly random.

FIG. 18 is a plot of the min-entropy that represents an embodiment of the present invention. In FIG. 18, horizontal axis 1802 corresponds to the probability Pr(x) of an event x, vertical axis 1804 represents values of the min-entropy, and curve 1806 represents the min-entropy H_min(X). When the maximum probability Pr(x) of the event x occurring is “1” 1808, the min-entropy is “0” 1810. In other words, the event x occurs with certainty and is completely deterministic. Note that an event with a probability Pr(x) equal to “0” also has a min-entropy of “0” 1810 and means that the event does not occur. When the maximum probability Pr(x) of the event x occurring is “½” 1812, the min-entropy is “1” 1814. In other words, the event x occurs without bias and corresponds to a truly random event. When the maximum probability of the event x occurring is greater than ½, the min-entropy is between “0” and “1” and is said to be a biased event, such as such as the point 1816, which corresponds to the min-entropy 1818.

In order to demonstrate use of the min-entropy, the following discussion describes how the elements of the density matrix are used in the definition of the min-entropy for three different kinds of ensembles of states generated by Eve. When Alice performs tomographic analysis on a single pulse in a pure state |ψ

provide by Eve, the random variable X is distributed over the set {0, 1}, and the min-entropy is:
H _Min(|ψ

ψ|)=−log₂(max(Pr _H(|ψ

),Pr _V(|ψ

)))
where

Pr_H(|ψ

)=|c|²=

H|ψ

², and

Pr_V(|ψ

)=|d|²=

V|ψ

²

The min-entropy can be extended to the case when Alice performs tomographic analysis on n pulses, all in the same pure state |ψ

provided by Eve. The random variable X is distributed over the set {0, 1}ⁿ, and the min-entropy is:
H _Min((|ψ

ψ|)ⁿ)=−n log₂(max(Pr _H(|ψ

),Pr _V(|ψ

)))
Finally, when Alice performs tomographic analysis on n pulses in a statistical mixture of pure states |ψ_i

provided by Eve, the min-entropy is:

$H_{\mathrm{Min}}\left({\left(\sum _i{p}_i{\psi }_i\rangle \langle {\psi }_i\right)}^n\right)=-n\sum _i{p}_i{\log }_2\left(\max \left({\mathrm{Pr}}_H\left({\psi }_i\rangle \right),{\mathrm{Pr}}_V\left({\psi }_i\rangle \right)\right)\right)$ $\mathrm{where}$ ${\mathrm{Pr}}_H\left({\psi }_i\rangle \right)=\sum _i{p}_i{c_i}^2,\mathrm{and}$ ${\mathrm{Pr}}_V\left({\psi }_i\rangle \right)=\sum _i{p}_i{d_i}^2$
Alice does not know the decomposition of the states comprising the pulse Eve is providing. Alice only has access to the density matrix {circumflex over (ρ)} that she generates during tomographic analysis. In order to obtain an extension of the min-entropy to arbitrary states, the min-entropy associated with pulses is defined as the minimum min-entropy over all the possible decompositions of the density matrix {circumflex over (ρ)}. Using such a definition of the minimum min-entropy places an upper bound on the amount of information Eve can obtain about Alice's sequence.

Note that as long as the min-entropy H_min is not equal zero, Eve does not have complete control over the sequences of bits produced by the QRBGs described above. In other words, as long as the min-entropy is greater than zero, there exists some number m of random bits within a sequence of n bits generated by the QRBGs, where m<n.

In order to facilitate the tomographic analysis, the min-entropy H_Min({circumflex over (ρ)}) is re-characterized as a function of the Stokes parameters. First, the 2×2 density matrix {circumflex over (ρ)} associated with a statistical mixture of states |ψ_i

above can be rewritten in terms of the Stokes parameters (S₀, S₁, S₂, S₃) as follows:

$\begin{array}{c}{\hat{\rho }}_S=\frac{1}{2}\sum _{i=0}^3\frac{S_i}{S_0}{\sigma }_i\\ =\frac{1}{2}\left[\begin{array}{cc}1+S_1& {\mathrm{<figure-callout\; id="2"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">S</figure-callout>}}_2+{\mathrm{iS}}_3\\ S_2-{\mathrm{iS}}_3& 1-S_1\end{array}\right]\end{array}$
where

subscript “S” identifies the density matrix rewritten in terms of the Stokes parameters;

the Stokes parameter S₀ is normalized to “1”; and

σ₁, σ₂, and σ₃ are the well-known Pauli matrices in the {|R

|L

basis.

The Stokes parameters of the density matrix {circumflex over (ρ)}_S can be determined based on the detection events as follows. In the device embodiments of the present invention described above with reference to FIGS. 11-14, Alice use the avalanche photodiodes 1132 and 1133 to detect single photons. The system control 1104 receives signals from the avalanche photodiodes 1132 and 1133 and calculates the average number of horizontally polarized photons,

H

and the average number of vertically polarized photons,

V

Alice uses the p-i-n photodetectors 1128-1131 to detect the intensities of electromagnetic radiation I(α₄₅), I(α₋₄₅), I(α_R), and I(α_L). The system control 1104 receives signals corresponding to the intensities and calculates the corresponding average intensities

α₄₅

,

α₋₄₅

α_R

and

α_L

The normalized Stokes parameters can then be determined by:

${\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1=\frac{\langle H\rangle -\langle V\rangle }{\langle H\rangle +\langle V\rangle },<figure-callout\; id="2"\; label="\; S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}"></figure-callout>S_{}{}_2=\frac{\langle {\alpha }_{45}\rangle -\langle {\alpha }_{-45}\rangle }{\langle {\mathrm{<figure-callout\; id="45"\; label="\alpha "\; filenames="US07849122-20101207-D00000.png,US07849122-20101207-D00005.png"\; state="\{\{state\}\}">\alpha </figure-callout>}}_{45}\rangle +\langle {\alpha }_{-45}\rangle },\mathrm{and}$ $S_3=\frac{\langle {\alpha }_R\rangle -\langle {\alpha }_L\rangle }{\langle {\alpha }_R\rangle +\langle {\alpha }_L\rangle }$

By defining the following real-valued function for all density matrices {circumflex over (ρ)}_S:

$f\left({\hat{\rho }}_S\right)=-{\log }_2\left(\frac{1+\sqrt{1-{{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="iS"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">iS</figure-callout>}}_2}^2}}{2}\right)$
the following theorem can be stated:

Theorem. The min-entropy of a system described by the density matrix {circumflex over (ρ)}_S is
H _Min({circumflex over (ρ)}_S)=∫({circumflex over (ρ)}_S)
Proof of the Theorem is provided below in an appendix. The Theorem demonstrates that a measurement of the density matrix of the states used to generate a sequence of bits has an upper bound on the amount of information an adversary, such as Eve, can obtain. Barak et al. shows that given a sequence of n bits with a min-entropy H_min, one can extract m random bits from the raw sequence of bits, where m<n. The m random bits are distributed according to a distribution which is arbitrarily close to a uniform distribution of bits.

FIG. 19 shows a control-flow diagram that represents one of many embodiments of the present invention for generating sequences of random bits. In step 1902, the QRBGs 1100 described above are employed N times to filter the polarization components |α₄₅

|α₋₄₅

|α_R

and |α_L

and generate and filter single photons in the polarization states |H

and |V

In step 1904, based on the detection results obtained in step 1902, the system control 1104 calculates the average intensities

α₄₅

α₋₄₅

α_R

and

α_L

which are used to determine the Stokes parameters S₂ and S₃, as described above. In step 1906, the routine “generate a sequence of raw bits” is called, which generates a raw sequence of n random bits {right arrow over (r)}_n from a sequence of N raw counts, as described above with reference to FIG. 16. In step 1908, the routine “tomographic analysis” is called. The routine tomographic analysis is a method for determining a density matrix {circumflex over (ρ)}_S and the min-entropy H_min({circumflex over (ρ)}_S), as described above. In step 1910, the routine “sift the sequence of raw bits” is called, which employs the min-entropy H_min, to remove bias from the sequence {right arrow over (r)}_n and produce a smaller sequence of m random bits {right arrow over (s)}_m. In step 1912, the sequence of random bits {right arrow over (s)}_m is output.

FIG. 20 shows a control-flow diagram for the routine “generate a sequence of raw bits” called in step 1906 of FIG. 19 and represents one of many embodiments of the present invention. In step 2002, N raw counts are collected as described above with reference to FIG. 15. In step 2004, the raw count is sifted by discarding the raw counts that correspond to either “No Pulse” or “Errors” leaving a raw sequence of n random bits, as described above with reference to FIG. 15. In step 2006, the system control 1104 averages of the raw counts corresponding to the states |H

and |V

in order to obtain determine the averages

H

and

V

which are subsequently used to determine the Stokes parameter S₁, as described above.

FIG. 21 shows a control-flow diagram for the routine “tomographic analysis” called in step 1908 of FIG. 19 and represents one of many embodiments of the present invention. In step 2102, the averages

α₄₅

α₋₄₅

α_R

α_L

, obtained in step 1904 of FIG. 19, and the averages

H

and

V

obtained in step 2006 of FIG. 20, are acquired. In step 2104, the density matrix {circumflex over (ρ)}_S is constructed as described above. In step 2106, the density matrix {circumflex over (ρ)}_S is used to construct the min-entropy H_min({circumflex over (ρ)}_S).

FIG. 22 is a control-flow diagram for the routine “sift the sequence of raw bits” called in step 1910 of FIG. 19 and represents one of many embodiments of the present invention. In step 2202, the raw sequence of random bits {right arrow over (r)}_n generated in the routine “generate sequence of raw binary numbers” in FIG. 20 is input. In step 2204, the min-entropy H_Min({circumflex over (ρ)}_S) generated in the routine “tomographic analysis” of FIG. 21 is input. In step 2206, the Toeplitz matrix T_m×n is constructed, as described in the Barak et al. reference. In step 2208, the sequence {right arrow over (s)}_m is determined and output in step, as described above with reference to FIG. 16.

Based on the Barak et al. reference, the maximum number of bits that can be extracted from a raw sequence of n ransom bits is:

$m=\mathrm{kn}-4{\mathrm{<figure-callout\; id="2"\; label="log"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">log</figure-callout>}}_2\left(\frac{1}{\varepsilon }\right)-2$
where ∈ is a statistical distance between a distribution of m bits and a uniform distribution. The statistical distance is defined mathematically as:

$\varepsilon =\frac{1}{2}\sum _a\mathrm{Pr}\left(X=a\right)-\mathrm{Pr}\left(W=a\right)$
where X and W denote different distributions. The yield Y is the fraction of random bits m/n that can be obtained from a raw sequence of random bits.

In other embodiments of the present invention, those skilled in the art would recognize that other optical quantum systems, states other than coherent states, such as thermal states, can be used. For example, the state generator 1002 may be a light bulb. In other embodiment of the present invention, polarization states other than the polarization states |H

|V

|α₄₅

|α₋₄₅

|α_R

and |α_L

can be used, such as the polarization states described in the James et al. reference.

The foregoing description, for purposes of explanation, used specific nomenclature to provide a thorough understanding of the invention. However, it will be apparent to one skilled in the art that the specific details are not required in order to practice the invention. The foregoing descriptions of specific embodiments of the present invention are presented for purposes of illustration and description. They are not intended to be exhaustive of or to limit the invention to the precise forms disclosed. Obviously, many modifications and variations are possible in view of the above teachings. The embodiments are shown and described in order to best explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the following claims and their equivalents:

APPENDIX

Theorem. The min-entropy of a system described by a density matrix {circumflex over (ρ)}_S is:
H _Min({circumflex over (ρ)}_S)=∫({circumflex over (ρ)}_S)
In order to demonstrate a proof of the Theorem, proofs of the followings three Lemmas are demonstrated.

Lemma 1. For each pure state |ψ

H _Min(|ψ

ψ|)=∫(|ψ

ψ|)
Proof of Lemma 1 is demonstrated by showing that:

$\max \left({\mathrm{Pr}}_H,{\mathrm{Pr}}_V\right)=\frac{1+\sqrt{1-{{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="iS"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">iS</figure-callout>}}_2}^2}}{2}$
for the cases Pr_H>½, Pr_H<½, and Pr_V=½. First, because |ψ

is a pure state, the associated Stokes parameters correspond to a point on the surface of a Poincare sphere, as described above with reference to FIG. 10, and, in particular, the parameters S₁ and S₂ are given by:
S ₁=√{square root over (4P _H(1−P _H))}cos 2ψ
S ₂=√{square root over (4P _H(1−P _H))}sin 2ψ
Substituting S₁ and S₂ into the right-hand side gives:

$\begin{array}{c}\frac{1+\sqrt{1-{{\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="iS"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">iS</figure-callout>}}_2}^2}}{2}=\frac{1+\sqrt{1-4{\mathrm{Pr}}_H\left(1-{\mathrm{Pr}}_H\right)}}{2}\\ =\frac{1}{2}+\frac{1}{2}-{\mathrm{Pr}}_H\end{array}$
When Pr_H>½ the left-hand side reduces to
max(Pr _H,1−Pr _H)=Pr _H,
and the right-hand side reduces to

$\frac{1}{2}+\frac{1}{2}-{\mathrm{Pr}}_H=\frac{1}{2}+{\mathrm{Pr}}_H-\frac{1}{2}={\mathrm{Pr}}_H$
When Pr_H<½ the left-hand side reduces to
max(Pr _H,1−Pr _H)=1−Pr _H
and the right-hand side reduces to

$\frac{1}{2}+\frac{1}{2}-{\mathrm{Pr}}_H=\frac{1}{2}+\frac{1}{2}-{\mathrm{Pr}}_H=1-{\mathrm{Pr}}_H$
Finally, for the trivial case, when Pr_HV=½, both the left and right hand side reduce to ½.

Lemma 2. Two pure states |ψ₁

and |ψ₂

that are represented by the density matrices:

${\hat{\rho }}_1=\frac{1}{2}\left[\begin{array}{cc}1+S_3^{\prime }& S_1-{\mathrm{iS}}_2\\ {\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="iS"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">iS</figure-callout>}}_2& 1-S_3^{\prime }\end{array}\right],\mathrm{and}$ ${\hat{\rho }}_2=\frac{1}{2}\left[\begin{array}{cc}1-S_3^{\prime }& S_1-{\mathrm{iS}}_2\\ {\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="iS"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">iS</figure-callout>}}_2& 1+S_3^{\prime }\end{array}\right]$
with S₃′=√{square root over (1−S₁ ²−S₂ ²)} are a decomposition of the density matrix

${\hat{\rho }}_S=\frac{1}{2}\left[\begin{array}{cc}1+S_3& S_1-{\mathrm{iS}}_2\\ {\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">S</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="iS"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">iS</figure-callout>}}_2& 1-S_3\end{array}\right]$
Proof of Lemma 2: The density matrices represent pure states that are a decomposition of {circumflex over (ρ)}_S with diagorial matrix elements that satisfy:

${\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">p</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">p</figure-callout>}}_2=1,\mathrm{and}$ $p_1-{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">p</figure-callout>}}_2=\frac{S_3}{S_3^{\prime }}$
Based on Lemma 1, since both |ψ₁

and |ψ₂

are pure states:
H _Min(|ψ₁

ψ₁|)=∫({circumflex over (ρ)}_S)=H _Min(|ψ₂

ψ₂|)
In addition, based on the equation for

$H_{\mathrm{Min}}\left({\left(\sum _i{p}_i{\psi }_i\rangle \langle {\psi }_i\right)}^n\right)$
above, with n=1:

$H_{\mathrm{Min}}\left({\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">p</figure-callout>}}_1{\mathrm{<figure-callout\; id="1"\; label="\psi "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">\psi </figure-callout>}}_1\rangle \langle {\psi }_1+{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">p</figure-callout>}}_2{\mathrm{<figure-callout\; id="2"\; label="\psi "\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">\psi </figure-callout>}}_2\rangle \langle {\psi }_2\right)={\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">p</figure-callout>}}_1{H}_{\mathrm{Min}}\left({\psi }_1\rangle \langle {\psi }_1\right)+{\mathrm{<figure-callout\; id="2"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00008.png"\; state="\{\{state\}\}">p</figure-callout>}}_2{H}_{\mathrm{Min}}\left({\psi }_2\rangle \langle {\psi }_2\right)=\left({\mathrm{<figure-callout\; id="1"\; label="p"\; filenames="US07849122-20101207-D00001.png,US07849122-20101207-D00003.png"\; state="\{\{state\}\}">p</figure-callout>}}_1+p_2\right)f\left({\hat{\rho }}_S\right)=f\left({\hat{\rho }}_S\right)$

Lemma 3. The function ∫({circumflex over (ρ)}_S) is a convex function of the Stokes parameters S₁, S₂, S₃ on the Poincare sphere.

Proof of Lemma 3: The eigenvalues of the Hessian matrix of ∫({circumflex over (ρ)}_S) are non-negative over the domain (½,1).

Proof of the Theorem. According to properties of a convex function:

$f\left({\hat{\rho }}_S\right)\le \sum _i{p}_{i}f\left({\psi }_i\rangle \langle {\psi }_i\right)$
for each decomposition of {circumflex over (ρ)}_S. Substituting the result of Lemma 1 and using equation

$H_{\mathrm{Min}}\left({\left(\sum _i{p}_i{\psi }_i\rangle \langle {\psi }_i\right)}^n\right)$
above gives:

$f\left({\hat{\rho }}_S\right)\le H_{\mathrm{Min}}\left(\sum _i{p}_i{\psi }_i\rangle \langle {\psi }_i\right)$
Which means that ∫({circumflex over (ρ)}_S) is a lower bound for the min-entropy of {circumflex over (ρ)}_S. But according to Lemma 2, there is at least one decomposition of {circumflex over (ρ)}_S for which
∫({circumflex over (ρ)}_S)=H _Min({circumflex over (ρ)}_S).
Therefore, ∫({circumflex over (ρ)}_S) is equal to the minimum of H_Min over all the decompositions of {circumflex over (ρ)}_S. Q.E.D.

Claims (20)

1. A self-authenticating, quantum random bit generator comprising:

a transmission layer including,

an electromagnetic radiation source coupled to a waveguide branching into a first waveguide, a second waveguide, and a third waveguide, the electromagnetic radiation source configured to generate pulses of electromagnetic radiation in a first polarization state;

two or more polarization rotators positioned and configured to rotate pulses transmitted in the second waveguide into a second polarization state and rotate pulses transmitted in the third waveguide into a third polarization state; and

a system control configured to generate a sequence of bits based on polarization basis states of the pulses transmitted in the first waveguide, and tomographically authenticate randomness of the sequence of bits based on polarization basis states of the pulses transmitted in the second and third waveguides.

2. The generator of claim 1 further comprises an attenuator operably coupled to the first waveguide and configured to reduce the energy of the pulses transmitted in the first waveguide to at most a single photon.

3. The generator of claim 1 further comprises a first polarizing beamsplitter coupled to the first waveguide, a second polarizing beamsplitter coupled to the second waveguide, and a third polarizing beamspitter coupled to the third waveguide.

4. The generator of claim 3 wherein the first, second, and third polarizing beamsplitters split each pulse into two polarization basis states.

5. The generator of claim 1 wherein the electromagnetic radiation source further comprises one of:

a diode laser; and

a light-emitting diode coupled to a waveguide polarizer and a polarization rotator.

6. The generator of claim 1 wherein each polarization rotator further comprise one of:

two or more polarization rotators;

a half-wave plate; and

a quarter-wave plate.

7. The generator of claim 1 wherein the system control further comprises:

a pair of avalanche photodiodes configured to detect photons transmitted in the first waveguide;

a first pair of p-i-n photodetectors configured to detect pulses transmitted in the second waveguide; and

a second pair of p-i-n photodetectors configured to detect pulses transmitted in the third waveguide.

8. The generator of claim 1 wherein the system control further comprises detectors configured to detect polarization basis states of the pulses transmitting in the first, second, and third waveguides.

9. The generator of claim 1 wherein the waveguides further comprise one of

ridge waveguides;

waveguides in a photonic crystal; and

optical fibers.

10. The generator of claim 1 wherein the transmission layer further comprises one of:

silicon oxynitride; and

an optical polymer.

11. An optoelectronic chip including the self-authenticating quantum random bit generator of claim 1 embedded within a layer of the optoelectronic chip.

12. A method for generating a sequence of random bits, the method comprising:

generating a sequence of electromagnetic radiation pulses, each pulse in a first polarization state;

splitting each pulse into a first pulse, a second pulse, and a third pulse, all three pulses in the same first polarization state;

rotating each second pulse into a second polarization state and each third pulse into a third polarization state using polarization rotators;

generating a sequence of bits based on detecting one of two polarization basis states of the first pulse; and

based on the polarization states of the second and third pulses, performing tomographic analysis in order to authenticate the randomness of the sequence of bits.

13. The method of claim 12 further comprising attenuating the first pulse intensity into a single photon.

14. The method of claim 12 further comprising splitting the first pulse into two polarization basis states, the second pulse into two polarization basis states, and the third pulse into two polarization basis states.

15. The method of claim 14 wherein splitting the pulses into two polarization basis states further comprises transmitting each pulse through a polarizing beamsplitter.

16. The method of claim 12 wherein rotating each second pulse into a second polarization state and each third pulse into a third polarization state further comprises transmitting each second pulse through one or more polarizing beamsplitters and transmitting each third pulse through one or more polarizing beamsplitters.

17. The method of claim 12 wherein generating a sequence of bits further comprises assigning a bit “1” to detecting one of the two polarization basis states of the first pulse and assigning a bit “0” to detecting the other of the two polarization basis states of the first pulse.

18. The method of claim 12 wherein performing tomographic analysis further comprises constructing the minimum entropy:

$H_{\mathrm{Min}}\left({\hat{\rho }}_S\right)\equiv -{\log }_2\left(\underset{x\in {\hat{\rho }}_S}{\max }\mathrm{Pr}\left(x\right)\right)$

where

{circumflex over (ρ)}_S is the density matrix for an ensemble of states |104 _i

=c_i|H

+d_i|V

as a function of Stokes parameters;

Pr (x) is the probability of a event x; and

$\underset{x\in X}{\max }\mathrm{Pr}\left(x\right)$

means the maximum probability Pr (x) over every event x in X.

19. The method of claim 18 wherein the density matrix further comprises:

$\begin{array}{c}{\hat{\rho }}_S=\frac{1}{2}\sum _{i=0}^3\frac{S_i}{S_0}{\sigma }_i\\ =\frac{1}{2}\left[\begin{array}{cc}1+S_3& S_1-{\mathrm{iS}}_2\\ S_1+{\mathrm{iS}}_2& 1-S_3\end{array}\right]\end{array}$

where

(S₀, S₁, S₂, S₃) are Stokes parameters;

the Stokes parameter S₀ is normalized to “1”; and

σ₁, σ₂, and σ₃ are the Pauli matrices.

20. The method of claim 18 wherein performing tomographic analysis further comprises constructing a Toeplitz matrix T_m×n based on the minimum entropy H_min({circumflex over (ρ)}_S), where m is the number of random bits, n is the number of raw bits, and m<n.

Images